pulumi/pkg/resource/deploy/step.go

1150 lines
43 KiB
Go
Raw Normal View History

2018-05-22 21:43:36 +02:00
// Copyright 2016-2018, Pulumi Corporation.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package deploy
import (
"fmt"
"strings"
"github.com/pkg/errors"
"github.com/pulumi/pulumi/pkg/v2/resource/deploy/providers"
"github.com/pulumi/pulumi/sdk/v2/go/common/diag"
"github.com/pulumi/pulumi/sdk/v2/go/common/diag/colors"
"github.com/pulumi/pulumi/sdk/v2/go/common/resource"
"github.com/pulumi/pulumi/sdk/v2/go/common/resource/plugin"
"github.com/pulumi/pulumi/sdk/v2/go/common/tokens"
"github.com/pulumi/pulumi/sdk/v2/go/common/util/contract"
"github.com/pulumi/pulumi/sdk/v2/go/common/util/logging"
)
// StepCompleteFunc is the type of functions returned from Step.Apply. These functions are to be called
// when the engine has fully retired a step.
type StepCompleteFunc func()
// Step is a specification for a deployment operation.
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
type Step interface {
// Apply applies or previews this step. It returns the status of the resource after the step application,
// a function to call to signal that this step has fully completed, and an error, if one occurred while applying
// the step.
//
// The returned StepCompleteFunc, if not nil, must be called after committing the results of this step into
// the state of the deployment.
Apply(preview bool) (resource.Status, StepCompleteFunc, error) // applies or previews this step.
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
Op() StepOp // the operation performed by this step.
URN() resource.URN // the resource URN (for before and after).
Type() tokens.Type // the type affected by this step.
Provider() string // the provider reference for this step.
Old() *resource.State // the state of the resource before performing this step.
New() *resource.State // the state of the resource after performing this step.
Res() *resource.State // the latest state for the resource that is known (worst case, old).
Logical() bool // true if this step represents a logical operation in the program.
Deployment() *Deployment // the owning deployment.
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
// SameStep is a mutating step that does nothing.
type SameStep struct {
deployment *Deployment // the current deployment.
reg RegisterResourceEvent // the registration intent to convey a URN back to.
old *resource.State // the state of the resource before this step.
new *resource.State // the state of the resource after this step.
// If this is a same-step for a resource being created but which was not --target'ed by the user
// (and thus was skipped).
skippedCreate bool
}
Switch to parent pointers; display components nicely This change switches from child lists to parent pointers, in the way resource ancestries are represented. This cleans up a fair bit of the old parenting logic, including all notion of ambient parent scopes (and will notably address pulumi/pulumi#435). This lets us show a more parent/child display in the output when doing planning and updating. For instance, here is an update of a lambda's text, which is logically part of a cloud timer: * cloud:timer:Timer: (same) [urn=urn:pulumi:malta::lm-cloud::cloud:timer:Timer::lm-cts-malta-job-CleanSnapshots] * cloud:function:Function: (same) [urn=urn:pulumi:malta::lm-cloud::cloud:function:Function::lm-cts-malta-job-CleanSnapshots] * aws:serverless:Function: (same) [urn=urn:pulumi:malta::lm-cloud::aws:serverless:Function::lm-cts-malta-job-CleanSnapshots] ~ aws:lambda/function:Function: (modify) [id=lm-cts-malta-job-CleanSnapshots-fee4f3bf41280741] [urn=urn:pulumi:malta::lm-cloud::aws:lambda/function:Function::lm-cts-malta-job-CleanSnapshots] - code : archive(assets:2092f44) { // etc etc etc Note that we still get walls of text, but this will be actually quite nice when combined with pulumi/pulumi#454. I've also suppressed printing properties that didn't change during updates when --detailed was not passed, and also suppressed empty strings and zero-length arrays (since TF uses these as defaults in many places and it just makes creation and deletion quite verbose). Note that this is a far cry from everything we can possibly do here as part of pulumi/pulumi#340 (and even pulumi/pulumi#417). But it's a good start towards taming some of our output spew.
2017-11-17 03:21:41 +01:00
var _ Step = (*SameStep)(nil)
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
func NewSameStep(deployment *Deployment, reg RegisterResourceEvent, old, new *resource.State) Step {
contract.Assert(old != nil)
contract.Assert(old.URN != "")
contract.Assert(old.ID != "" || !old.Custom)
Implement first-class providers. (#1695) ### First-Class Providers These changes implement support for first-class providers. First-class providers are provider plugins that are exposed as resources via the Pulumi programming model so that they may be explicitly and multiply instantiated. Each instance of a provider resource may be configured differently, and configuration parameters may be source from the outputs of other resources. ### Provider Plugin Changes In order to accommodate the need to verify and diff provider configuration and configure providers without complete configuration information, these changes adjust the high-level provider plugin interface. Two new methods for validating a provider's configuration and diffing changes to the same have been added (`CheckConfig` and `DiffConfig`, respectively), and the type of the configuration bag accepted by `Configure` has been changed to a `PropertyMap`. These changes have not yet been reflected in the provider plugin gRPC interface. We will do this in a set of follow-up changes. Until then, these methods are implemented by adapters: - `CheckConfig` validates that all configuration parameters are string or unknown properties. This is necessary because existing plugins only accept string-typed configuration values. - `DiffConfig` either returns "never replace" if all configuration values are known or "must replace" if any configuration value is unknown. The justification for this behavior is given [here](https://github.com/pulumi/pulumi/pull/1695/files#diff-a6cd5c7f337665f5bb22e92ca5f07537R106) - `Configure` converts the config bag to a legacy config map and configures the provider plugin if all config values are known. If any config value is unknown, the underlying plugin is not configured and the provider may only perform `Check`, `Read`, and `Invoke`, all of which return empty results. We justify this behavior becuase it is only possible during a preview and provides the best experience we can manage with the existing gRPC interface. ### Resource Model Changes Providers are now exposed as resources that participate in a stack's dependency graph. Like other resources, they are explicitly created, may have multiple instances, and may have dependencies on other resources. Providers are referred to using provider references, which are a combination of the provider's URN and its ID. This design addresses the need during a preview to refer to providers that have not yet been physically created and therefore have no ID. All custom resources that are not themselves providers must specify a single provider via a provider reference. The named provider will be used to manage that resource's CRUD operations. If a resource's provider reference changes, the resource must be replaced. Though its URN is not present in the resource's dependency list, the provider should be treated as a dependency of the resource when topologically sorting the dependency graph. Finally, `Invoke` operations must now specify a provider to use for the invocation via a provider reference. ### Engine Changes First-class providers support requires a few changes to the engine: - The engine must have some way to map from provider references to provider plugins. It must be possible to add providers from a stack's checkpoint to this map and to register new/updated providers during the execution of a plan in response to CRUD operations on provider resources. - In order to support updating existing stacks using existing Pulumi programs that may not explicitly instantiate providers, the engine must be able to manage the "default" providers for each package referenced by a checkpoint or Pulumi program. The configuration for a "default" provider is taken from the stack's configuration data. The former need is addressed by adding a provider registry type that is responsible for managing all of the plugins required by a plan. In addition to loading plugins froma checkpoint and providing the ability to map from a provider reference to a provider plugin, this type serves as the provider plugin for providers themselves (i.e. it is the "provider provider"). The latter need is solved via two relatively self-contained changes to plan setup and the eval source. During plan setup, the old checkpoint is scanned for custom resources that do not have a provider reference in order to compute the set of packages that require a default provider. Once this set has been computed, the required default provider definitions are conjured and prepended to the checkpoint's resource list. Each resource that requires a default provider is then updated to refer to the default provider for its package. While an eval source is running, each custom resource registration, resource read, and invoke that does not name a provider is trapped before being returned by the source iterator. If no default provider for the appropriate package has been registered, the eval source synthesizes an appropriate registration, waits for it to complete, and records the registered provider's reference. This reference is injected into the original request, which is then processed as usual. If a default provider was already registered, the recorded reference is used and no new registration occurs. ### SDK Changes These changes only expose first-class providers from the Node.JS SDK. - A new abstract class, `ProviderResource`, can be subclassed and used to instantiate first-class providers. - A new field in `ResourceOptions`, `provider`, can be used to supply a particular provider instance to manage a `CustomResource`'s CRUD operations. - A new type, `InvokeOptions`, can be used to specify options that control the behavior of a call to `pulumi.runtime.invoke`. This type includes a `provider` field that is analogous to `ResourceOptions.provider`.
2018-08-07 02:50:29 +02:00
contract.Assert(!old.Custom || old.Provider != "" || providers.IsProviderType(old.Type))
contract.Assert(!old.Delete)
contract.Assert(new != nil)
contract.Assert(new.URN != "")
contract.Assert(new.ID == "")
contract.Assert(!new.Custom || new.Provider != "" || providers.IsProviderType(new.Type))
contract.Assert(!new.Delete)
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
return &SameStep{
deployment: deployment,
reg: reg,
old: old,
new: new,
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
}
// NewSkippedCreateStep produces a SameStep for a resource that was created but not targeted
// by the user (and thus was skipped). These act as no-op steps (hence 'same') since we are not
// actually creating the resource, but ensure that we complete resource-registration and convey the
// right information downstream. For example, we will not write these into the checkpoint file.
func NewSkippedCreateStep(deployment *Deployment, reg RegisterResourceEvent, new *resource.State) Step {
contract.Assert(new != nil)
contract.Assert(new.URN != "")
contract.Assert(new.ID == "")
contract.Assert(!new.Custom || new.Provider != "" || providers.IsProviderType(new.Type))
contract.Assert(!new.Delete)
// Make the old state here a direct copy of the new state
old := *new
return &SameStep{
deployment: deployment,
reg: reg,
old: &old,
new: new,
skippedCreate: true,
}
}
func (s *SameStep) Op() StepOp { return OpSame }
func (s *SameStep) Deployment() *Deployment { return s.deployment }
func (s *SameStep) Type() tokens.Type { return s.new.Type }
func (s *SameStep) Provider() string { return s.new.Provider }
func (s *SameStep) URN() resource.URN { return s.new.URN }
func (s *SameStep) Old() *resource.State { return s.old }
func (s *SameStep) New() *resource.State { return s.new }
func (s *SameStep) Res() *resource.State { return s.new }
func (s *SameStep) Logical() bool { return true }
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
func (s *SameStep) Apply(preview bool) (resource.Status, StepCompleteFunc, error) {
// Retain the ID, and outputs:
s.new.ID = s.old.ID
s.new.Outputs = s.old.Outputs
Propagate inputs to outputs during preview. (#3327) These changes restore a more-correct version of the behavior that was disabled with #3014. The original implementation of this behavior was done in the SDKs, which do not have access to the complete inputs for a resource (in particular, default values filled in by the provider during `Check` are not exposed to the SDK). This lack of information meant that the resolved output values could disagree with the typings present in a provider SDK. Exacerbating this problem was the fact that unknown values were dropped entirely, causing `undefined` values to appear in unexpected places. By doing this in the engine and allowing unknown values to be represented in a first-class manner in the SDK, we can attack both of these issues. Although this behavior is not _strictly_ consistent with respect to the resource model--in an update, a resource's output properties will come from its provider and may differ from its input properties--this behavior was present in the product for a fairly long time without significant issues. In the future, we may be able to improve the accuracy of resource outputs during a preview by allowing the provider to dry-run CRUD operations and return partially-known values where possible. These changes also introduce new APIs in the Node and Python SDKs that work with unknown values in a first-class fashion: - A new parameter to the `apply` function that indicates that the callback should be run even if the result of the apply contains unknown values - `containsUnknowns` and `isUnknown`, which return true if a value either contains nested unknown values or is exactly an unknown value - The `Unknown` type, which represents unknown values The primary use case for these APIs is to allow nested, properties with known values to be accessed via the lifted property accessor even when the containing property is not fully know. A common example of this pattern is the `metadata.name` property of a Kubernetes `Namespace` object: while other properties of the `metadata` bag may be unknown, `name` is often known. These APIs allow `ns.metadata.name` to return a known value in this case. In order to avoid exposing downlevel SDKs to unknown values--a change which could break user code by exposing it to unexpected values--a language SDK must indicate whether or not it supports first-class unknown values as part of each `RegisterResourceRequest`. These changes also allow us to avoid breaking user code with the new behavior introduced by the prior commit. Fixes #3190.
2019-11-11 21:09:34 +01:00
complete := func() { s.reg.Done(&RegisterResult{State: s.new}) }
return resource.StatusOK, complete, nil
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
func (s *SameStep) IsSkippedCreate() bool {
return s.skippedCreate
}
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
// CreateStep is a mutating step that creates an entirely new resource.
type CreateStep struct {
deployment *Deployment // the current deployment.
Defer all diffs to resource providers. (#2849) Thse changes make a subtle but critical adjustment to the process the Pulumi engine uses to determine whether or not a difference exists between a resource's actual and desired states, and adjusts the way this difference is calculated and displayed accordingly. Today, the Pulumi engine get the first chance to decide whether or not there is a difference between a resource's actual and desired states. It does this by comparing the current set of inputs for a resource (i.e. the inputs from the running Pulumi program) with the last set of inputs used to update the resource. If there is no difference between the old and new inputs, the engine decides that no change is necessary without consulting the resource's provider. Only if there are changes does the engine consult the resource's provider for more information about the difference. This can be problematic for a number of reasons: - Not all providers do input-input comparison; some do input-state comparison - Not all providers are able to update the last deployed set of inputs when performing a refresh - Some providers--either intentionally or due to bugs--may see changes in resources whose inputs have not changed All of these situations are confusing at the very least, and the first is problematic with respect to correctness. Furthermore, the display code only renders diffs it observes rather than rendering the diffs observed by the provider, which can obscure the actual changes detected at runtime. These changes address both of these issues: - Rather than comparing the current inputs against the last inputs before calling a resource provider's Diff function, the engine calls the Diff function in all cases. - Providers may now return a list of properties that differ between the requested and actual state and the way in which they differ. This information will then be used by the CLI to render the diff appropriately. A provider may also indicate that a particular diff is between old and new inputs rather than old state and new inputs. Fixes #2453.
2019-07-01 21:34:19 +02:00
reg RegisterResourceEvent // the registration intent to convey a URN back to.
old *resource.State // the state of the existing resource (only for replacements).
new *resource.State // the state of the resource after this step.
keys []resource.PropertyKey // the keys causing replacement (only for replacements).
diffs []resource.PropertyKey // the keys causing a diff (only for replacements).
detailedDiff map[string]plugin.PropertyDiff // the structured property diff (only for replacements).
replacing bool // true if this is a create due to a replacement.
pendingDelete bool // true if this replacement should create a pending delete.
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
Switch to parent pointers; display components nicely This change switches from child lists to parent pointers, in the way resource ancestries are represented. This cleans up a fair bit of the old parenting logic, including all notion of ambient parent scopes (and will notably address pulumi/pulumi#435). This lets us show a more parent/child display in the output when doing planning and updating. For instance, here is an update of a lambda's text, which is logically part of a cloud timer: * cloud:timer:Timer: (same) [urn=urn:pulumi:malta::lm-cloud::cloud:timer:Timer::lm-cts-malta-job-CleanSnapshots] * cloud:function:Function: (same) [urn=urn:pulumi:malta::lm-cloud::cloud:function:Function::lm-cts-malta-job-CleanSnapshots] * aws:serverless:Function: (same) [urn=urn:pulumi:malta::lm-cloud::aws:serverless:Function::lm-cts-malta-job-CleanSnapshots] ~ aws:lambda/function:Function: (modify) [id=lm-cts-malta-job-CleanSnapshots-fee4f3bf41280741] [urn=urn:pulumi:malta::lm-cloud::aws:lambda/function:Function::lm-cts-malta-job-CleanSnapshots] - code : archive(assets:2092f44) { // etc etc etc Note that we still get walls of text, but this will be actually quite nice when combined with pulumi/pulumi#454. I've also suppressed printing properties that didn't change during updates when --detailed was not passed, and also suppressed empty strings and zero-length arrays (since TF uses these as defaults in many places and it just makes creation and deletion quite verbose). Note that this is a far cry from everything we can possibly do here as part of pulumi/pulumi#340 (and even pulumi/pulumi#417). But it's a good start towards taming some of our output spew.
2017-11-17 03:21:41 +01:00
var _ Step = (*CreateStep)(nil)
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
func NewCreateStep(deployment *Deployment, reg RegisterResourceEvent, new *resource.State) Step {
contract.Assert(reg != nil)
contract.Assert(new != nil)
contract.Assert(new.URN != "")
contract.Assert(new.ID == "")
Implement first-class providers. (#1695) ### First-Class Providers These changes implement support for first-class providers. First-class providers are provider plugins that are exposed as resources via the Pulumi programming model so that they may be explicitly and multiply instantiated. Each instance of a provider resource may be configured differently, and configuration parameters may be source from the outputs of other resources. ### Provider Plugin Changes In order to accommodate the need to verify and diff provider configuration and configure providers without complete configuration information, these changes adjust the high-level provider plugin interface. Two new methods for validating a provider's configuration and diffing changes to the same have been added (`CheckConfig` and `DiffConfig`, respectively), and the type of the configuration bag accepted by `Configure` has been changed to a `PropertyMap`. These changes have not yet been reflected in the provider plugin gRPC interface. We will do this in a set of follow-up changes. Until then, these methods are implemented by adapters: - `CheckConfig` validates that all configuration parameters are string or unknown properties. This is necessary because existing plugins only accept string-typed configuration values. - `DiffConfig` either returns "never replace" if all configuration values are known or "must replace" if any configuration value is unknown. The justification for this behavior is given [here](https://github.com/pulumi/pulumi/pull/1695/files#diff-a6cd5c7f337665f5bb22e92ca5f07537R106) - `Configure` converts the config bag to a legacy config map and configures the provider plugin if all config values are known. If any config value is unknown, the underlying plugin is not configured and the provider may only perform `Check`, `Read`, and `Invoke`, all of which return empty results. We justify this behavior becuase it is only possible during a preview and provides the best experience we can manage with the existing gRPC interface. ### Resource Model Changes Providers are now exposed as resources that participate in a stack's dependency graph. Like other resources, they are explicitly created, may have multiple instances, and may have dependencies on other resources. Providers are referred to using provider references, which are a combination of the provider's URN and its ID. This design addresses the need during a preview to refer to providers that have not yet been physically created and therefore have no ID. All custom resources that are not themselves providers must specify a single provider via a provider reference. The named provider will be used to manage that resource's CRUD operations. If a resource's provider reference changes, the resource must be replaced. Though its URN is not present in the resource's dependency list, the provider should be treated as a dependency of the resource when topologically sorting the dependency graph. Finally, `Invoke` operations must now specify a provider to use for the invocation via a provider reference. ### Engine Changes First-class providers support requires a few changes to the engine: - The engine must have some way to map from provider references to provider plugins. It must be possible to add providers from a stack's checkpoint to this map and to register new/updated providers during the execution of a plan in response to CRUD operations on provider resources. - In order to support updating existing stacks using existing Pulumi programs that may not explicitly instantiate providers, the engine must be able to manage the "default" providers for each package referenced by a checkpoint or Pulumi program. The configuration for a "default" provider is taken from the stack's configuration data. The former need is addressed by adding a provider registry type that is responsible for managing all of the plugins required by a plan. In addition to loading plugins froma checkpoint and providing the ability to map from a provider reference to a provider plugin, this type serves as the provider plugin for providers themselves (i.e. it is the "provider provider"). The latter need is solved via two relatively self-contained changes to plan setup and the eval source. During plan setup, the old checkpoint is scanned for custom resources that do not have a provider reference in order to compute the set of packages that require a default provider. Once this set has been computed, the required default provider definitions are conjured and prepended to the checkpoint's resource list. Each resource that requires a default provider is then updated to refer to the default provider for its package. While an eval source is running, each custom resource registration, resource read, and invoke that does not name a provider is trapped before being returned by the source iterator. If no default provider for the appropriate package has been registered, the eval source synthesizes an appropriate registration, waits for it to complete, and records the registered provider's reference. This reference is injected into the original request, which is then processed as usual. If a default provider was already registered, the recorded reference is used and no new registration occurs. ### SDK Changes These changes only expose first-class providers from the Node.JS SDK. - A new abstract class, `ProviderResource`, can be subclassed and used to instantiate first-class providers. - A new field in `ResourceOptions`, `provider`, can be used to supply a particular provider instance to manage a `CustomResource`'s CRUD operations. - A new type, `InvokeOptions`, can be used to specify options that control the behavior of a call to `pulumi.runtime.invoke`. This type includes a `provider` field that is analogous to `ResourceOptions.provider`.
2018-08-07 02:50:29 +02:00
contract.Assert(!new.Custom || new.Provider != "" || providers.IsProviderType(new.Type))
contract.Assert(!new.Delete)
contract.Assert(!new.External)
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
return &CreateStep{
deployment: deployment,
reg: reg,
new: new,
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
}
func NewCreateReplacementStep(deployment *Deployment, reg RegisterResourceEvent, old, new *resource.State,
Defer all diffs to resource providers. (#2849) Thse changes make a subtle but critical adjustment to the process the Pulumi engine uses to determine whether or not a difference exists between a resource's actual and desired states, and adjusts the way this difference is calculated and displayed accordingly. Today, the Pulumi engine get the first chance to decide whether or not there is a difference between a resource's actual and desired states. It does this by comparing the current set of inputs for a resource (i.e. the inputs from the running Pulumi program) with the last set of inputs used to update the resource. If there is no difference between the old and new inputs, the engine decides that no change is necessary without consulting the resource's provider. Only if there are changes does the engine consult the resource's provider for more information about the difference. This can be problematic for a number of reasons: - Not all providers do input-input comparison; some do input-state comparison - Not all providers are able to update the last deployed set of inputs when performing a refresh - Some providers--either intentionally or due to bugs--may see changes in resources whose inputs have not changed All of these situations are confusing at the very least, and the first is problematic with respect to correctness. Furthermore, the display code only renders diffs it observes rather than rendering the diffs observed by the provider, which can obscure the actual changes detected at runtime. These changes address both of these issues: - Rather than comparing the current inputs against the last inputs before calling a resource provider's Diff function, the engine calls the Diff function in all cases. - Providers may now return a list of properties that differ between the requested and actual state and the way in which they differ. This information will then be used by the CLI to render the diff appropriately. A provider may also indicate that a particular diff is between old and new inputs rather than old state and new inputs. Fixes #2453.
2019-07-01 21:34:19 +02:00
keys, diffs []resource.PropertyKey, detailedDiff map[string]plugin.PropertyDiff, pendingDelete bool) Step {
contract.Assert(reg != nil)
contract.Assert(old != nil)
contract.Assert(old.URN != "")
contract.Assert(old.ID != "" || !old.Custom)
contract.Assert(!old.Delete)
contract.Assert(new != nil)
contract.Assert(new.URN != "")
contract.Assert(new.ID == "")
Implement first-class providers. (#1695) ### First-Class Providers These changes implement support for first-class providers. First-class providers are provider plugins that are exposed as resources via the Pulumi programming model so that they may be explicitly and multiply instantiated. Each instance of a provider resource may be configured differently, and configuration parameters may be source from the outputs of other resources. ### Provider Plugin Changes In order to accommodate the need to verify and diff provider configuration and configure providers without complete configuration information, these changes adjust the high-level provider plugin interface. Two new methods for validating a provider's configuration and diffing changes to the same have been added (`CheckConfig` and `DiffConfig`, respectively), and the type of the configuration bag accepted by `Configure` has been changed to a `PropertyMap`. These changes have not yet been reflected in the provider plugin gRPC interface. We will do this in a set of follow-up changes. Until then, these methods are implemented by adapters: - `CheckConfig` validates that all configuration parameters are string or unknown properties. This is necessary because existing plugins only accept string-typed configuration values. - `DiffConfig` either returns "never replace" if all configuration values are known or "must replace" if any configuration value is unknown. The justification for this behavior is given [here](https://github.com/pulumi/pulumi/pull/1695/files#diff-a6cd5c7f337665f5bb22e92ca5f07537R106) - `Configure` converts the config bag to a legacy config map and configures the provider plugin if all config values are known. If any config value is unknown, the underlying plugin is not configured and the provider may only perform `Check`, `Read`, and `Invoke`, all of which return empty results. We justify this behavior becuase it is only possible during a preview and provides the best experience we can manage with the existing gRPC interface. ### Resource Model Changes Providers are now exposed as resources that participate in a stack's dependency graph. Like other resources, they are explicitly created, may have multiple instances, and may have dependencies on other resources. Providers are referred to using provider references, which are a combination of the provider's URN and its ID. This design addresses the need during a preview to refer to providers that have not yet been physically created and therefore have no ID. All custom resources that are not themselves providers must specify a single provider via a provider reference. The named provider will be used to manage that resource's CRUD operations. If a resource's provider reference changes, the resource must be replaced. Though its URN is not present in the resource's dependency list, the provider should be treated as a dependency of the resource when topologically sorting the dependency graph. Finally, `Invoke` operations must now specify a provider to use for the invocation via a provider reference. ### Engine Changes First-class providers support requires a few changes to the engine: - The engine must have some way to map from provider references to provider plugins. It must be possible to add providers from a stack's checkpoint to this map and to register new/updated providers during the execution of a plan in response to CRUD operations on provider resources. - In order to support updating existing stacks using existing Pulumi programs that may not explicitly instantiate providers, the engine must be able to manage the "default" providers for each package referenced by a checkpoint or Pulumi program. The configuration for a "default" provider is taken from the stack's configuration data. The former need is addressed by adding a provider registry type that is responsible for managing all of the plugins required by a plan. In addition to loading plugins froma checkpoint and providing the ability to map from a provider reference to a provider plugin, this type serves as the provider plugin for providers themselves (i.e. it is the "provider provider"). The latter need is solved via two relatively self-contained changes to plan setup and the eval source. During plan setup, the old checkpoint is scanned for custom resources that do not have a provider reference in order to compute the set of packages that require a default provider. Once this set has been computed, the required default provider definitions are conjured and prepended to the checkpoint's resource list. Each resource that requires a default provider is then updated to refer to the default provider for its package. While an eval source is running, each custom resource registration, resource read, and invoke that does not name a provider is trapped before being returned by the source iterator. If no default provider for the appropriate package has been registered, the eval source synthesizes an appropriate registration, waits for it to complete, and records the registered provider's reference. This reference is injected into the original request, which is then processed as usual. If a default provider was already registered, the recorded reference is used and no new registration occurs. ### SDK Changes These changes only expose first-class providers from the Node.JS SDK. - A new abstract class, `ProviderResource`, can be subclassed and used to instantiate first-class providers. - A new field in `ResourceOptions`, `provider`, can be used to supply a particular provider instance to manage a `CustomResource`'s CRUD operations. - A new type, `InvokeOptions`, can be used to specify options that control the behavior of a call to `pulumi.runtime.invoke`. This type includes a `provider` field that is analogous to `ResourceOptions.provider`.
2018-08-07 02:50:29 +02:00
contract.Assert(!new.Custom || new.Provider != "" || providers.IsProviderType(new.Type))
contract.Assert(!new.Delete)
contract.Assert(!new.External)
return &CreateStep{
deployment: deployment,
reg: reg,
old: old,
new: new,
keys: keys,
diffs: diffs,
Defer all diffs to resource providers. (#2849) Thse changes make a subtle but critical adjustment to the process the Pulumi engine uses to determine whether or not a difference exists between a resource's actual and desired states, and adjusts the way this difference is calculated and displayed accordingly. Today, the Pulumi engine get the first chance to decide whether or not there is a difference between a resource's actual and desired states. It does this by comparing the current set of inputs for a resource (i.e. the inputs from the running Pulumi program) with the last set of inputs used to update the resource. If there is no difference between the old and new inputs, the engine decides that no change is necessary without consulting the resource's provider. Only if there are changes does the engine consult the resource's provider for more information about the difference. This can be problematic for a number of reasons: - Not all providers do input-input comparison; some do input-state comparison - Not all providers are able to update the last deployed set of inputs when performing a refresh - Some providers--either intentionally or due to bugs--may see changes in resources whose inputs have not changed All of these situations are confusing at the very least, and the first is problematic with respect to correctness. Furthermore, the display code only renders diffs it observes rather than rendering the diffs observed by the provider, which can obscure the actual changes detected at runtime. These changes address both of these issues: - Rather than comparing the current inputs against the last inputs before calling a resource provider's Diff function, the engine calls the Diff function in all cases. - Providers may now return a list of properties that differ between the requested and actual state and the way in which they differ. This information will then be used by the CLI to render the diff appropriately. A provider may also indicate that a particular diff is between old and new inputs rather than old state and new inputs. Fixes #2453.
2019-07-01 21:34:19 +02:00
detailedDiff: detailedDiff,
replacing: true,
pendingDelete: pendingDelete,
}
}
func (s *CreateStep) Op() StepOp {
if s.replacing {
return OpCreateReplacement
}
return OpCreate
}
func (s *CreateStep) Deployment() *Deployment { return s.deployment }
func (s *CreateStep) Type() tokens.Type { return s.new.Type }
Defer all diffs to resource providers. (#2849) Thse changes make a subtle but critical adjustment to the process the Pulumi engine uses to determine whether or not a difference exists between a resource's actual and desired states, and adjusts the way this difference is calculated and displayed accordingly. Today, the Pulumi engine get the first chance to decide whether or not there is a difference between a resource's actual and desired states. It does this by comparing the current set of inputs for a resource (i.e. the inputs from the running Pulumi program) with the last set of inputs used to update the resource. If there is no difference between the old and new inputs, the engine decides that no change is necessary without consulting the resource's provider. Only if there are changes does the engine consult the resource's provider for more information about the difference. This can be problematic for a number of reasons: - Not all providers do input-input comparison; some do input-state comparison - Not all providers are able to update the last deployed set of inputs when performing a refresh - Some providers--either intentionally or due to bugs--may see changes in resources whose inputs have not changed All of these situations are confusing at the very least, and the first is problematic with respect to correctness. Furthermore, the display code only renders diffs it observes rather than rendering the diffs observed by the provider, which can obscure the actual changes detected at runtime. These changes address both of these issues: - Rather than comparing the current inputs against the last inputs before calling a resource provider's Diff function, the engine calls the Diff function in all cases. - Providers may now return a list of properties that differ between the requested and actual state and the way in which they differ. This information will then be used by the CLI to render the diff appropriately. A provider may also indicate that a particular diff is between old and new inputs rather than old state and new inputs. Fixes #2453.
2019-07-01 21:34:19 +02:00
func (s *CreateStep) Provider() string { return s.new.Provider }
func (s *CreateStep) URN() resource.URN { return s.new.URN }
Defer all diffs to resource providers. (#2849) Thse changes make a subtle but critical adjustment to the process the Pulumi engine uses to determine whether or not a difference exists between a resource's actual and desired states, and adjusts the way this difference is calculated and displayed accordingly. Today, the Pulumi engine get the first chance to decide whether or not there is a difference between a resource's actual and desired states. It does this by comparing the current set of inputs for a resource (i.e. the inputs from the running Pulumi program) with the last set of inputs used to update the resource. If there is no difference between the old and new inputs, the engine decides that no change is necessary without consulting the resource's provider. Only if there are changes does the engine consult the resource's provider for more information about the difference. This can be problematic for a number of reasons: - Not all providers do input-input comparison; some do input-state comparison - Not all providers are able to update the last deployed set of inputs when performing a refresh - Some providers--either intentionally or due to bugs--may see changes in resources whose inputs have not changed All of these situations are confusing at the very least, and the first is problematic with respect to correctness. Furthermore, the display code only renders diffs it observes rather than rendering the diffs observed by the provider, which can obscure the actual changes detected at runtime. These changes address both of these issues: - Rather than comparing the current inputs against the last inputs before calling a resource provider's Diff function, the engine calls the Diff function in all cases. - Providers may now return a list of properties that differ between the requested and actual state and the way in which they differ. This information will then be used by the CLI to render the diff appropriately. A provider may also indicate that a particular diff is between old and new inputs rather than old state and new inputs. Fixes #2453.
2019-07-01 21:34:19 +02:00
func (s *CreateStep) Old() *resource.State { return s.old }
func (s *CreateStep) New() *resource.State { return s.new }
func (s *CreateStep) Res() *resource.State { return s.new }
func (s *CreateStep) Keys() []resource.PropertyKey { return s.keys }
func (s *CreateStep) Diffs() []resource.PropertyKey { return s.diffs }
func (s *CreateStep) DetailedDiff() map[string]plugin.PropertyDiff { return s.detailedDiff }
func (s *CreateStep) Logical() bool { return !s.replacing }
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
func (s *CreateStep) Apply(preview bool) (resource.Status, StepCompleteFunc, error) {
var resourceError error
resourceStatus := resource.StatusOK
if s.new.Custom {
// Invoke the Create RPC function for this provider:
prov, err := getProvider(s)
if err != nil {
return resource.StatusOK, nil, err
}
2019-07-15 23:26:28 +02:00
id, outs, rst, err := prov.Create(s.URN(), s.new.Inputs, s.new.CustomTimeouts.Create, s.deployment.preview)
if err != nil {
if rst != resource.StatusPartialFailure {
return rst, nil, err
}
resourceError = err
resourceStatus = rst
if initErr, isInitErr := err.(*plugin.InitError); isInitErr {
s.new.InitErrors = initErr.Reasons
}
}
if !preview && id == "" {
return resourceStatus, nil, fmt.Errorf("provider did not return an ID from Create")
Implement components This change implements core support for "components" in the Pulumi Fabric. This work is described further in pulumi/pulumi#340, where we are still discussing some of the finer points. In a nutshell, resources no longer imply external providers. It's entirely possible to have a resource that logically represents something but without having a physical manifestation that needs to be tracked and managed by our typical CRUD operations. For example, the aws/serverless/Function helper is one such type. It aggregates Lambda-related resources and exposes a nice interface. All of the Pulumi Cloud Framework resources are also examples. To indicate that a resource does participate in the usual CRUD resource provider, it simply derives from ExternalResource instead of Resource. All resources now have the ability to adopt children. This is purely a metadata/tagging thing, and will help us roll up displays, provide attribution to the developer, and even hide aspects of the resource graph as appropriate (e.g., when they are implementation details). Our use of this capability is ultra limited right now; in fact, the only place we display children is in the CLI output. For instance: + aws:serverless:Function: (create) [urn=urn:pulumi:demo::serverless::aws:serverless:Function::mylambda] => urn:pulumi:demo::serverless::aws:iam/role:Role::mylambda-iamrole => urn:pulumi:demo::serverless::aws:iam/rolePolicyAttachment:RolePolicyAttachment::mylambda-iampolicy-0 => urn:pulumi:demo::serverless::aws:lambda/function:Function::mylambda The bit indicating whether a resource is external or not is tracked in the resulting checkpoint file, along with any of its children.
2017-10-14 23:18:43 +02:00
}
// Copy any of the default and output properties on the live object state.
s.new.ID = id
s.new.Outputs = outs
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
// Mark the old resource as pending deletion if necessary.
if s.replacing && s.pendingDelete {
s.old.Delete = true
}
complete := func() { s.reg.Done(&RegisterResult{State: s.new}) }
2018-07-07 00:17:32 +02:00
if resourceError == nil {
return resourceStatus, complete, nil
2018-07-07 00:17:32 +02:00
}
return resourceStatus, complete, resourceError
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
// DeleteStep is a mutating step that deletes an existing resource. If `old` is marked "External",
// DeleteStep is a no-op.
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
type DeleteStep struct {
deployment *Deployment // the current deployment.
old *resource.State // the state of the existing resource.
replacing bool // true if part of a replacement.
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
Switch to parent pointers; display components nicely This change switches from child lists to parent pointers, in the way resource ancestries are represented. This cleans up a fair bit of the old parenting logic, including all notion of ambient parent scopes (and will notably address pulumi/pulumi#435). This lets us show a more parent/child display in the output when doing planning and updating. For instance, here is an update of a lambda's text, which is logically part of a cloud timer: * cloud:timer:Timer: (same) [urn=urn:pulumi:malta::lm-cloud::cloud:timer:Timer::lm-cts-malta-job-CleanSnapshots] * cloud:function:Function: (same) [urn=urn:pulumi:malta::lm-cloud::cloud:function:Function::lm-cts-malta-job-CleanSnapshots] * aws:serverless:Function: (same) [urn=urn:pulumi:malta::lm-cloud::aws:serverless:Function::lm-cts-malta-job-CleanSnapshots] ~ aws:lambda/function:Function: (modify) [id=lm-cts-malta-job-CleanSnapshots-fee4f3bf41280741] [urn=urn:pulumi:malta::lm-cloud::aws:lambda/function:Function::lm-cts-malta-job-CleanSnapshots] - code : archive(assets:2092f44) { // etc etc etc Note that we still get walls of text, but this will be actually quite nice when combined with pulumi/pulumi#454. I've also suppressed printing properties that didn't change during updates when --detailed was not passed, and also suppressed empty strings and zero-length arrays (since TF uses these as defaults in many places and it just makes creation and deletion quite verbose). Note that this is a far cry from everything we can possibly do here as part of pulumi/pulumi#340 (and even pulumi/pulumi#417). But it's a good start towards taming some of our output spew.
2017-11-17 03:21:41 +01:00
var _ Step = (*DeleteStep)(nil)
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
func NewDeleteStep(deployment *Deployment, old *resource.State) Step {
contract.Assert(old != nil)
contract.Assert(old.URN != "")
contract.Assert(old.ID != "" || !old.Custom)
Implement first-class providers. (#1695) ### First-Class Providers These changes implement support for first-class providers. First-class providers are provider plugins that are exposed as resources via the Pulumi programming model so that they may be explicitly and multiply instantiated. Each instance of a provider resource may be configured differently, and configuration parameters may be source from the outputs of other resources. ### Provider Plugin Changes In order to accommodate the need to verify and diff provider configuration and configure providers without complete configuration information, these changes adjust the high-level provider plugin interface. Two new methods for validating a provider's configuration and diffing changes to the same have been added (`CheckConfig` and `DiffConfig`, respectively), and the type of the configuration bag accepted by `Configure` has been changed to a `PropertyMap`. These changes have not yet been reflected in the provider plugin gRPC interface. We will do this in a set of follow-up changes. Until then, these methods are implemented by adapters: - `CheckConfig` validates that all configuration parameters are string or unknown properties. This is necessary because existing plugins only accept string-typed configuration values. - `DiffConfig` either returns "never replace" if all configuration values are known or "must replace" if any configuration value is unknown. The justification for this behavior is given [here](https://github.com/pulumi/pulumi/pull/1695/files#diff-a6cd5c7f337665f5bb22e92ca5f07537R106) - `Configure` converts the config bag to a legacy config map and configures the provider plugin if all config values are known. If any config value is unknown, the underlying plugin is not configured and the provider may only perform `Check`, `Read`, and `Invoke`, all of which return empty results. We justify this behavior becuase it is only possible during a preview and provides the best experience we can manage with the existing gRPC interface. ### Resource Model Changes Providers are now exposed as resources that participate in a stack's dependency graph. Like other resources, they are explicitly created, may have multiple instances, and may have dependencies on other resources. Providers are referred to using provider references, which are a combination of the provider's URN and its ID. This design addresses the need during a preview to refer to providers that have not yet been physically created and therefore have no ID. All custom resources that are not themselves providers must specify a single provider via a provider reference. The named provider will be used to manage that resource's CRUD operations. If a resource's provider reference changes, the resource must be replaced. Though its URN is not present in the resource's dependency list, the provider should be treated as a dependency of the resource when topologically sorting the dependency graph. Finally, `Invoke` operations must now specify a provider to use for the invocation via a provider reference. ### Engine Changes First-class providers support requires a few changes to the engine: - The engine must have some way to map from provider references to provider plugins. It must be possible to add providers from a stack's checkpoint to this map and to register new/updated providers during the execution of a plan in response to CRUD operations on provider resources. - In order to support updating existing stacks using existing Pulumi programs that may not explicitly instantiate providers, the engine must be able to manage the "default" providers for each package referenced by a checkpoint or Pulumi program. The configuration for a "default" provider is taken from the stack's configuration data. The former need is addressed by adding a provider registry type that is responsible for managing all of the plugins required by a plan. In addition to loading plugins froma checkpoint and providing the ability to map from a provider reference to a provider plugin, this type serves as the provider plugin for providers themselves (i.e. it is the "provider provider"). The latter need is solved via two relatively self-contained changes to plan setup and the eval source. During plan setup, the old checkpoint is scanned for custom resources that do not have a provider reference in order to compute the set of packages that require a default provider. Once this set has been computed, the required default provider definitions are conjured and prepended to the checkpoint's resource list. Each resource that requires a default provider is then updated to refer to the default provider for its package. While an eval source is running, each custom resource registration, resource read, and invoke that does not name a provider is trapped before being returned by the source iterator. If no default provider for the appropriate package has been registered, the eval source synthesizes an appropriate registration, waits for it to complete, and records the registered provider's reference. This reference is injected into the original request, which is then processed as usual. If a default provider was already registered, the recorded reference is used and no new registration occurs. ### SDK Changes These changes only expose first-class providers from the Node.JS SDK. - A new abstract class, `ProviderResource`, can be subclassed and used to instantiate first-class providers. - A new field in `ResourceOptions`, `provider`, can be used to supply a particular provider instance to manage a `CustomResource`'s CRUD operations. - A new type, `InvokeOptions`, can be used to specify options that control the behavior of a call to `pulumi.runtime.invoke`. This type includes a `provider` field that is analogous to `ResourceOptions.provider`.
2018-08-07 02:50:29 +02:00
contract.Assert(!old.Custom || old.Provider != "" || providers.IsProviderType(old.Type))
return &DeleteStep{
deployment: deployment,
old: old,
}
}
func NewDeleteReplacementStep(deployment *Deployment, old *resource.State, pendingReplace bool) Step {
contract.Assert(old != nil)
contract.Assert(old.URN != "")
contract.Assert(old.ID != "" || !old.Custom)
Implement first-class providers. (#1695) ### First-Class Providers These changes implement support for first-class providers. First-class providers are provider plugins that are exposed as resources via the Pulumi programming model so that they may be explicitly and multiply instantiated. Each instance of a provider resource may be configured differently, and configuration parameters may be source from the outputs of other resources. ### Provider Plugin Changes In order to accommodate the need to verify and diff provider configuration and configure providers without complete configuration information, these changes adjust the high-level provider plugin interface. Two new methods for validating a provider's configuration and diffing changes to the same have been added (`CheckConfig` and `DiffConfig`, respectively), and the type of the configuration bag accepted by `Configure` has been changed to a `PropertyMap`. These changes have not yet been reflected in the provider plugin gRPC interface. We will do this in a set of follow-up changes. Until then, these methods are implemented by adapters: - `CheckConfig` validates that all configuration parameters are string or unknown properties. This is necessary because existing plugins only accept string-typed configuration values. - `DiffConfig` either returns "never replace" if all configuration values are known or "must replace" if any configuration value is unknown. The justification for this behavior is given [here](https://github.com/pulumi/pulumi/pull/1695/files#diff-a6cd5c7f337665f5bb22e92ca5f07537R106) - `Configure` converts the config bag to a legacy config map and configures the provider plugin if all config values are known. If any config value is unknown, the underlying plugin is not configured and the provider may only perform `Check`, `Read`, and `Invoke`, all of which return empty results. We justify this behavior becuase it is only possible during a preview and provides the best experience we can manage with the existing gRPC interface. ### Resource Model Changes Providers are now exposed as resources that participate in a stack's dependency graph. Like other resources, they are explicitly created, may have multiple instances, and may have dependencies on other resources. Providers are referred to using provider references, which are a combination of the provider's URN and its ID. This design addresses the need during a preview to refer to providers that have not yet been physically created and therefore have no ID. All custom resources that are not themselves providers must specify a single provider via a provider reference. The named provider will be used to manage that resource's CRUD operations. If a resource's provider reference changes, the resource must be replaced. Though its URN is not present in the resource's dependency list, the provider should be treated as a dependency of the resource when topologically sorting the dependency graph. Finally, `Invoke` operations must now specify a provider to use for the invocation via a provider reference. ### Engine Changes First-class providers support requires a few changes to the engine: - The engine must have some way to map from provider references to provider plugins. It must be possible to add providers from a stack's checkpoint to this map and to register new/updated providers during the execution of a plan in response to CRUD operations on provider resources. - In order to support updating existing stacks using existing Pulumi programs that may not explicitly instantiate providers, the engine must be able to manage the "default" providers for each package referenced by a checkpoint or Pulumi program. The configuration for a "default" provider is taken from the stack's configuration data. The former need is addressed by adding a provider registry type that is responsible for managing all of the plugins required by a plan. In addition to loading plugins froma checkpoint and providing the ability to map from a provider reference to a provider plugin, this type serves as the provider plugin for providers themselves (i.e. it is the "provider provider"). The latter need is solved via two relatively self-contained changes to plan setup and the eval source. During plan setup, the old checkpoint is scanned for custom resources that do not have a provider reference in order to compute the set of packages that require a default provider. Once this set has been computed, the required default provider definitions are conjured and prepended to the checkpoint's resource list. Each resource that requires a default provider is then updated to refer to the default provider for its package. While an eval source is running, each custom resource registration, resource read, and invoke that does not name a provider is trapped before being returned by the source iterator. If no default provider for the appropriate package has been registered, the eval source synthesizes an appropriate registration, waits for it to complete, and records the registered provider's reference. This reference is injected into the original request, which is then processed as usual. If a default provider was already registered, the recorded reference is used and no new registration occurs. ### SDK Changes These changes only expose first-class providers from the Node.JS SDK. - A new abstract class, `ProviderResource`, can be subclassed and used to instantiate first-class providers. - A new field in `ResourceOptions`, `provider`, can be used to supply a particular provider instance to manage a `CustomResource`'s CRUD operations. - A new type, `InvokeOptions`, can be used to specify options that control the behavior of a call to `pulumi.runtime.invoke`. This type includes a `provider` field that is analogous to `ResourceOptions.provider`.
2018-08-07 02:50:29 +02:00
contract.Assert(!old.Custom || old.Provider != "" || providers.IsProviderType(old.Type))
Implement more precise delete-before-replace semantics. (#2369) This implements the new algorithm for deciding which resources must be deleted due to a delete-before-replace operation. We need to compute the set of resources that may be replaced by a change to the resource under consideration. We do this by taking the complete set of transitive dependents on the resource under consideration and removing any resources that would not be replaced by changes to their dependencies. We determine whether or not a resource may be replaced by substituting unknowns for input properties that may change due to deletion of the resources their value depends on and calling the resource provider's Diff method. This is perhaps clearer when described by example. Consider the following dependency graph: A __|__ B C | _|_ D E F In this graph, all of B, C, D, E, and F transitively depend on A. It may be the case, however, that changes to the specific properties of any of those resources R that would occur if a resource on the path to A were deleted and recreated may not cause R to be replaced. For example, the edge from B to A may be a simple dependsOn edge such that a change to B does not actually influence any of B's input properties. In that case, neither B nor D would need to be deleted before A could be deleted. In order to make the above algorithm a reality, the resource monitor interface has been updated to include a map that associates an input property key with the list of resources that input property depends on. Older clients of the resource monitor will leave this map empty, in which case all input properties will be treated as depending on all dependencies of the resource. This is probably overly conservative, but it is less conservative than what we currently implement, and is certainly correct.
2019-01-28 18:46:30 +01:00
// There are two cases in which we create a delete-replacment step:
//
// 1. When creating the delete steps that occur due to a delete-before-replace
// 2. When creating the delete step that occurs due to a delete-after-replace
//
// In the former case, the persistence layer may require that the resource remain in the
// checkpoint file for purposes of checkpoint integrity. We communicate this case by means
// of the `PendingReplacement` field on `resource.State`, which we set here.
//
// In the latter case, the resource must be deleted, but the deletion may not occur if an earlier step fails.
// The engine requires that the fact that the old resource must be deleted is persisted in the checkpoint so
// that it can issue a deletion of this resource on the next update to this stack.
contract.Assert(pendingReplace != old.Delete)
old.PendingReplacement = pendingReplace
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
return &DeleteStep{
deployment: deployment,
old: old,
replacing: true,
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
}
func (s *DeleteStep) Op() StepOp {
if s.old.External {
if s.replacing {
return OpDiscardReplaced
}
return OpReadDiscard
}
if s.replacing {
return OpDeleteReplaced
}
return OpDelete
}
func (s *DeleteStep) Deployment() *Deployment { return s.deployment }
func (s *DeleteStep) Type() tokens.Type { return s.old.Type }
func (s *DeleteStep) Provider() string { return s.old.Provider }
func (s *DeleteStep) URN() resource.URN { return s.old.URN }
func (s *DeleteStep) Old() *resource.State { return s.old }
func (s *DeleteStep) New() *resource.State { return nil }
func (s *DeleteStep) Res() *resource.State { return s.old }
func (s *DeleteStep) Logical() bool { return !s.replacing }
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
func (s *DeleteStep) Apply(preview bool) (resource.Status, StepCompleteFunc, error) {
// Refuse to delete protected resources.
if s.old.Protect {
return resource.StatusOK, nil,
errors.Errorf("refusing to delete protected resource '%s'", s.old.URN)
}
// Deleting an External resource is a no-op, since Pulumi does not own the lifecycle.
if !preview && !s.old.External {
if s.old.Custom {
// Invoke the Delete RPC function for this provider:
prov, err := getProvider(s)
if err != nil {
return resource.StatusOK, nil, err
}
2019-07-15 23:26:28 +02:00
if rst, err := prov.Delete(s.URN(), s.old.ID, s.old.Outputs, s.old.CustomTimeouts.Delete); err != nil {
return rst, nil, err
}
Implement components This change implements core support for "components" in the Pulumi Fabric. This work is described further in pulumi/pulumi#340, where we are still discussing some of the finer points. In a nutshell, resources no longer imply external providers. It's entirely possible to have a resource that logically represents something but without having a physical manifestation that needs to be tracked and managed by our typical CRUD operations. For example, the aws/serverless/Function helper is one such type. It aggregates Lambda-related resources and exposes a nice interface. All of the Pulumi Cloud Framework resources are also examples. To indicate that a resource does participate in the usual CRUD resource provider, it simply derives from ExternalResource instead of Resource. All resources now have the ability to adopt children. This is purely a metadata/tagging thing, and will help us roll up displays, provide attribution to the developer, and even hide aspects of the resource graph as appropriate (e.g., when they are implementation details). Our use of this capability is ultra limited right now; in fact, the only place we display children is in the CLI output. For instance: + aws:serverless:Function: (create) [urn=urn:pulumi:demo::serverless::aws:serverless:Function::mylambda] => urn:pulumi:demo::serverless::aws:iam/role:Role::mylambda-iamrole => urn:pulumi:demo::serverless::aws:iam/rolePolicyAttachment:RolePolicyAttachment::mylambda-iampolicy-0 => urn:pulumi:demo::serverless::aws:lambda/function:Function::mylambda The bit indicating whether a resource is external or not is tracked in the resulting checkpoint file, along with any of its children.
2017-10-14 23:18:43 +02:00
}
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
return resource.StatusOK, func() {}, nil
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
Implement more precise delete-before-replace semantics. (#2369) This implements the new algorithm for deciding which resources must be deleted due to a delete-before-replace operation. We need to compute the set of resources that may be replaced by a change to the resource under consideration. We do this by taking the complete set of transitive dependents on the resource under consideration and removing any resources that would not be replaced by changes to their dependencies. We determine whether or not a resource may be replaced by substituting unknowns for input properties that may change due to deletion of the resources their value depends on and calling the resource provider's Diff method. This is perhaps clearer when described by example. Consider the following dependency graph: A __|__ B C | _|_ D E F In this graph, all of B, C, D, E, and F transitively depend on A. It may be the case, however, that changes to the specific properties of any of those resources R that would occur if a resource on the path to A were deleted and recreated may not cause R to be replaced. For example, the edge from B to A may be a simple dependsOn edge such that a change to B does not actually influence any of B's input properties. In that case, neither B nor D would need to be deleted before A could be deleted. In order to make the above algorithm a reality, the resource monitor interface has been updated to include a map that associates an input property key with the list of resources that input property depends on. Older clients of the resource monitor will leave this map empty, in which case all input properties will be treated as depending on all dependencies of the resource. This is probably overly conservative, but it is less conservative than what we currently implement, and is certainly correct.
2019-01-28 18:46:30 +01:00
type RemovePendingReplaceStep struct {
deployment *Deployment // the current deployment.
old *resource.State // the state of the existing resource.
Implement more precise delete-before-replace semantics. (#2369) This implements the new algorithm for deciding which resources must be deleted due to a delete-before-replace operation. We need to compute the set of resources that may be replaced by a change to the resource under consideration. We do this by taking the complete set of transitive dependents on the resource under consideration and removing any resources that would not be replaced by changes to their dependencies. We determine whether or not a resource may be replaced by substituting unknowns for input properties that may change due to deletion of the resources their value depends on and calling the resource provider's Diff method. This is perhaps clearer when described by example. Consider the following dependency graph: A __|__ B C | _|_ D E F In this graph, all of B, C, D, E, and F transitively depend on A. It may be the case, however, that changes to the specific properties of any of those resources R that would occur if a resource on the path to A were deleted and recreated may not cause R to be replaced. For example, the edge from B to A may be a simple dependsOn edge such that a change to B does not actually influence any of B's input properties. In that case, neither B nor D would need to be deleted before A could be deleted. In order to make the above algorithm a reality, the resource monitor interface has been updated to include a map that associates an input property key with the list of resources that input property depends on. Older clients of the resource monitor will leave this map empty, in which case all input properties will be treated as depending on all dependencies of the resource. This is probably overly conservative, but it is less conservative than what we currently implement, and is certainly correct.
2019-01-28 18:46:30 +01:00
}
func NewRemovePendingReplaceStep(deployment *Deployment, old *resource.State) Step {
Implement more precise delete-before-replace semantics. (#2369) This implements the new algorithm for deciding which resources must be deleted due to a delete-before-replace operation. We need to compute the set of resources that may be replaced by a change to the resource under consideration. We do this by taking the complete set of transitive dependents on the resource under consideration and removing any resources that would not be replaced by changes to their dependencies. We determine whether or not a resource may be replaced by substituting unknowns for input properties that may change due to deletion of the resources their value depends on and calling the resource provider's Diff method. This is perhaps clearer when described by example. Consider the following dependency graph: A __|__ B C | _|_ D E F In this graph, all of B, C, D, E, and F transitively depend on A. It may be the case, however, that changes to the specific properties of any of those resources R that would occur if a resource on the path to A were deleted and recreated may not cause R to be replaced. For example, the edge from B to A may be a simple dependsOn edge such that a change to B does not actually influence any of B's input properties. In that case, neither B nor D would need to be deleted before A could be deleted. In order to make the above algorithm a reality, the resource monitor interface has been updated to include a map that associates an input property key with the list of resources that input property depends on. Older clients of the resource monitor will leave this map empty, in which case all input properties will be treated as depending on all dependencies of the resource. This is probably overly conservative, but it is less conservative than what we currently implement, and is certainly correct.
2019-01-28 18:46:30 +01:00
contract.Assert(old != nil)
contract.Assert(old.PendingReplacement)
return &RemovePendingReplaceStep{
deployment: deployment,
old: old,
Implement more precise delete-before-replace semantics. (#2369) This implements the new algorithm for deciding which resources must be deleted due to a delete-before-replace operation. We need to compute the set of resources that may be replaced by a change to the resource under consideration. We do this by taking the complete set of transitive dependents on the resource under consideration and removing any resources that would not be replaced by changes to their dependencies. We determine whether or not a resource may be replaced by substituting unknowns for input properties that may change due to deletion of the resources their value depends on and calling the resource provider's Diff method. This is perhaps clearer when described by example. Consider the following dependency graph: A __|__ B C | _|_ D E F In this graph, all of B, C, D, E, and F transitively depend on A. It may be the case, however, that changes to the specific properties of any of those resources R that would occur if a resource on the path to A were deleted and recreated may not cause R to be replaced. For example, the edge from B to A may be a simple dependsOn edge such that a change to B does not actually influence any of B's input properties. In that case, neither B nor D would need to be deleted before A could be deleted. In order to make the above algorithm a reality, the resource monitor interface has been updated to include a map that associates an input property key with the list of resources that input property depends on. Older clients of the resource monitor will leave this map empty, in which case all input properties will be treated as depending on all dependencies of the resource. This is probably overly conservative, but it is less conservative than what we currently implement, and is certainly correct.
2019-01-28 18:46:30 +01:00
}
}
func (s *RemovePendingReplaceStep) Op() StepOp {
return OpRemovePendingReplace
}
func (s *RemovePendingReplaceStep) Deployment() *Deployment { return s.deployment }
func (s *RemovePendingReplaceStep) Type() tokens.Type { return s.old.Type }
func (s *RemovePendingReplaceStep) Provider() string { return s.old.Provider }
func (s *RemovePendingReplaceStep) URN() resource.URN { return s.old.URN }
func (s *RemovePendingReplaceStep) Old() *resource.State { return s.old }
func (s *RemovePendingReplaceStep) New() *resource.State { return nil }
func (s *RemovePendingReplaceStep) Res() *resource.State { return s.old }
func (s *RemovePendingReplaceStep) Logical() bool { return false }
Implement more precise delete-before-replace semantics. (#2369) This implements the new algorithm for deciding which resources must be deleted due to a delete-before-replace operation. We need to compute the set of resources that may be replaced by a change to the resource under consideration. We do this by taking the complete set of transitive dependents on the resource under consideration and removing any resources that would not be replaced by changes to their dependencies. We determine whether or not a resource may be replaced by substituting unknowns for input properties that may change due to deletion of the resources their value depends on and calling the resource provider's Diff method. This is perhaps clearer when described by example. Consider the following dependency graph: A __|__ B C | _|_ D E F In this graph, all of B, C, D, E, and F transitively depend on A. It may be the case, however, that changes to the specific properties of any of those resources R that would occur if a resource on the path to A were deleted and recreated may not cause R to be replaced. For example, the edge from B to A may be a simple dependsOn edge such that a change to B does not actually influence any of B's input properties. In that case, neither B nor D would need to be deleted before A could be deleted. In order to make the above algorithm a reality, the resource monitor interface has been updated to include a map that associates an input property key with the list of resources that input property depends on. Older clients of the resource monitor will leave this map empty, in which case all input properties will be treated as depending on all dependencies of the resource. This is probably overly conservative, but it is less conservative than what we currently implement, and is certainly correct.
2019-01-28 18:46:30 +01:00
func (s *RemovePendingReplaceStep) Apply(preview bool) (resource.Status, StepCompleteFunc, error) {
return resource.StatusOK, nil, nil
}
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
// UpdateStep is a mutating step that updates an existing resource's state.
type UpdateStep struct {
deployment *Deployment // the current deployment.
reg RegisterResourceEvent // the registration intent to convey a URN back to.
old *resource.State // the state of the existing resource.
new *resource.State // the newly computed state of the resource after updating.
stables []resource.PropertyKey // an optional list of properties that won't change during this update.
diffs []resource.PropertyKey // the keys causing a diff.
detailedDiff map[string]plugin.PropertyDiff // the structured diff.
ignoreChanges []string // a list of property paths to ignore when updating.
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
Switch to parent pointers; display components nicely This change switches from child lists to parent pointers, in the way resource ancestries are represented. This cleans up a fair bit of the old parenting logic, including all notion of ambient parent scopes (and will notably address pulumi/pulumi#435). This lets us show a more parent/child display in the output when doing planning and updating. For instance, here is an update of a lambda's text, which is logically part of a cloud timer: * cloud:timer:Timer: (same) [urn=urn:pulumi:malta::lm-cloud::cloud:timer:Timer::lm-cts-malta-job-CleanSnapshots] * cloud:function:Function: (same) [urn=urn:pulumi:malta::lm-cloud::cloud:function:Function::lm-cts-malta-job-CleanSnapshots] * aws:serverless:Function: (same) [urn=urn:pulumi:malta::lm-cloud::aws:serverless:Function::lm-cts-malta-job-CleanSnapshots] ~ aws:lambda/function:Function: (modify) [id=lm-cts-malta-job-CleanSnapshots-fee4f3bf41280741] [urn=urn:pulumi:malta::lm-cloud::aws:lambda/function:Function::lm-cts-malta-job-CleanSnapshots] - code : archive(assets:2092f44) { // etc etc etc Note that we still get walls of text, but this will be actually quite nice when combined with pulumi/pulumi#454. I've also suppressed printing properties that didn't change during updates when --detailed was not passed, and also suppressed empty strings and zero-length arrays (since TF uses these as defaults in many places and it just makes creation and deletion quite verbose). Note that this is a far cry from everything we can possibly do here as part of pulumi/pulumi#340 (and even pulumi/pulumi#417). But it's a good start towards taming some of our output spew.
2017-11-17 03:21:41 +01:00
var _ Step = (*UpdateStep)(nil)
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
func NewUpdateStep(deployment *Deployment, reg RegisterResourceEvent, old, new *resource.State,
stables, diffs []resource.PropertyKey, detailedDiff map[string]plugin.PropertyDiff,
ignoreChanges []string) Step {
contract.Assert(old != nil)
contract.Assert(old.URN != "")
contract.Assert(old.ID != "" || !old.Custom)
Implement first-class providers. (#1695) ### First-Class Providers These changes implement support for first-class providers. First-class providers are provider plugins that are exposed as resources via the Pulumi programming model so that they may be explicitly and multiply instantiated. Each instance of a provider resource may be configured differently, and configuration parameters may be source from the outputs of other resources. ### Provider Plugin Changes In order to accommodate the need to verify and diff provider configuration and configure providers without complete configuration information, these changes adjust the high-level provider plugin interface. Two new methods for validating a provider's configuration and diffing changes to the same have been added (`CheckConfig` and `DiffConfig`, respectively), and the type of the configuration bag accepted by `Configure` has been changed to a `PropertyMap`. These changes have not yet been reflected in the provider plugin gRPC interface. We will do this in a set of follow-up changes. Until then, these methods are implemented by adapters: - `CheckConfig` validates that all configuration parameters are string or unknown properties. This is necessary because existing plugins only accept string-typed configuration values. - `DiffConfig` either returns "never replace" if all configuration values are known or "must replace" if any configuration value is unknown. The justification for this behavior is given [here](https://github.com/pulumi/pulumi/pull/1695/files#diff-a6cd5c7f337665f5bb22e92ca5f07537R106) - `Configure` converts the config bag to a legacy config map and configures the provider plugin if all config values are known. If any config value is unknown, the underlying plugin is not configured and the provider may only perform `Check`, `Read`, and `Invoke`, all of which return empty results. We justify this behavior becuase it is only possible during a preview and provides the best experience we can manage with the existing gRPC interface. ### Resource Model Changes Providers are now exposed as resources that participate in a stack's dependency graph. Like other resources, they are explicitly created, may have multiple instances, and may have dependencies on other resources. Providers are referred to using provider references, which are a combination of the provider's URN and its ID. This design addresses the need during a preview to refer to providers that have not yet been physically created and therefore have no ID. All custom resources that are not themselves providers must specify a single provider via a provider reference. The named provider will be used to manage that resource's CRUD operations. If a resource's provider reference changes, the resource must be replaced. Though its URN is not present in the resource's dependency list, the provider should be treated as a dependency of the resource when topologically sorting the dependency graph. Finally, `Invoke` operations must now specify a provider to use for the invocation via a provider reference. ### Engine Changes First-class providers support requires a few changes to the engine: - The engine must have some way to map from provider references to provider plugins. It must be possible to add providers from a stack's checkpoint to this map and to register new/updated providers during the execution of a plan in response to CRUD operations on provider resources. - In order to support updating existing stacks using existing Pulumi programs that may not explicitly instantiate providers, the engine must be able to manage the "default" providers for each package referenced by a checkpoint or Pulumi program. The configuration for a "default" provider is taken from the stack's configuration data. The former need is addressed by adding a provider registry type that is responsible for managing all of the plugins required by a plan. In addition to loading plugins froma checkpoint and providing the ability to map from a provider reference to a provider plugin, this type serves as the provider plugin for providers themselves (i.e. it is the "provider provider"). The latter need is solved via two relatively self-contained changes to plan setup and the eval source. During plan setup, the old checkpoint is scanned for custom resources that do not have a provider reference in order to compute the set of packages that require a default provider. Once this set has been computed, the required default provider definitions are conjured and prepended to the checkpoint's resource list. Each resource that requires a default provider is then updated to refer to the default provider for its package. While an eval source is running, each custom resource registration, resource read, and invoke that does not name a provider is trapped before being returned by the source iterator. If no default provider for the appropriate package has been registered, the eval source synthesizes an appropriate registration, waits for it to complete, and records the registered provider's reference. This reference is injected into the original request, which is then processed as usual. If a default provider was already registered, the recorded reference is used and no new registration occurs. ### SDK Changes These changes only expose first-class providers from the Node.JS SDK. - A new abstract class, `ProviderResource`, can be subclassed and used to instantiate first-class providers. - A new field in `ResourceOptions`, `provider`, can be used to supply a particular provider instance to manage a `CustomResource`'s CRUD operations. - A new type, `InvokeOptions`, can be used to specify options that control the behavior of a call to `pulumi.runtime.invoke`. This type includes a `provider` field that is analogous to `ResourceOptions.provider`.
2018-08-07 02:50:29 +02:00
contract.Assert(!old.Custom || old.Provider != "" || providers.IsProviderType(old.Type))
contract.Assert(!old.Delete)
contract.Assert(new != nil)
contract.Assert(new.URN != "")
contract.Assert(new.ID == "")
contract.Assert(!new.Custom || new.Provider != "" || providers.IsProviderType(new.Type))
contract.Assert(!new.Delete)
contract.Assert(!new.External)
contract.Assert(!old.External)
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
return &UpdateStep{
deployment: deployment,
reg: reg,
old: old,
new: new,
stables: stables,
diffs: diffs,
detailedDiff: detailedDiff,
ignoreChanges: ignoreChanges,
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
}
Defer all diffs to resource providers. (#2849) Thse changes make a subtle but critical adjustment to the process the Pulumi engine uses to determine whether or not a difference exists between a resource's actual and desired states, and adjusts the way this difference is calculated and displayed accordingly. Today, the Pulumi engine get the first chance to decide whether or not there is a difference between a resource's actual and desired states. It does this by comparing the current set of inputs for a resource (i.e. the inputs from the running Pulumi program) with the last set of inputs used to update the resource. If there is no difference between the old and new inputs, the engine decides that no change is necessary without consulting the resource's provider. Only if there are changes does the engine consult the resource's provider for more information about the difference. This can be problematic for a number of reasons: - Not all providers do input-input comparison; some do input-state comparison - Not all providers are able to update the last deployed set of inputs when performing a refresh - Some providers--either intentionally or due to bugs--may see changes in resources whose inputs have not changed All of these situations are confusing at the very least, and the first is problematic with respect to correctness. Furthermore, the display code only renders diffs it observes rather than rendering the diffs observed by the provider, which can obscure the actual changes detected at runtime. These changes address both of these issues: - Rather than comparing the current inputs against the last inputs before calling a resource provider's Diff function, the engine calls the Diff function in all cases. - Providers may now return a list of properties that differ between the requested and actual state and the way in which they differ. This information will then be used by the CLI to render the diff appropriately. A provider may also indicate that a particular diff is between old and new inputs rather than old state and new inputs. Fixes #2453.
2019-07-01 21:34:19 +02:00
func (s *UpdateStep) Op() StepOp { return OpUpdate }
func (s *UpdateStep) Deployment() *Deployment { return s.deployment }
func (s *UpdateStep) Type() tokens.Type { return s.new.Type }
Defer all diffs to resource providers. (#2849) Thse changes make a subtle but critical adjustment to the process the Pulumi engine uses to determine whether or not a difference exists between a resource's actual and desired states, and adjusts the way this difference is calculated and displayed accordingly. Today, the Pulumi engine get the first chance to decide whether or not there is a difference between a resource's actual and desired states. It does this by comparing the current set of inputs for a resource (i.e. the inputs from the running Pulumi program) with the last set of inputs used to update the resource. If there is no difference between the old and new inputs, the engine decides that no change is necessary without consulting the resource's provider. Only if there are changes does the engine consult the resource's provider for more information about the difference. This can be problematic for a number of reasons: - Not all providers do input-input comparison; some do input-state comparison - Not all providers are able to update the last deployed set of inputs when performing a refresh - Some providers--either intentionally or due to bugs--may see changes in resources whose inputs have not changed All of these situations are confusing at the very least, and the first is problematic with respect to correctness. Furthermore, the display code only renders diffs it observes rather than rendering the diffs observed by the provider, which can obscure the actual changes detected at runtime. These changes address both of these issues: - Rather than comparing the current inputs against the last inputs before calling a resource provider's Diff function, the engine calls the Diff function in all cases. - Providers may now return a list of properties that differ between the requested and actual state and the way in which they differ. This information will then be used by the CLI to render the diff appropriately. A provider may also indicate that a particular diff is between old and new inputs rather than old state and new inputs. Fixes #2453.
2019-07-01 21:34:19 +02:00
func (s *UpdateStep) Provider() string { return s.new.Provider }
func (s *UpdateStep) URN() resource.URN { return s.new.URN }
Defer all diffs to resource providers. (#2849) Thse changes make a subtle but critical adjustment to the process the Pulumi engine uses to determine whether or not a difference exists between a resource's actual and desired states, and adjusts the way this difference is calculated and displayed accordingly. Today, the Pulumi engine get the first chance to decide whether or not there is a difference between a resource's actual and desired states. It does this by comparing the current set of inputs for a resource (i.e. the inputs from the running Pulumi program) with the last set of inputs used to update the resource. If there is no difference between the old and new inputs, the engine decides that no change is necessary without consulting the resource's provider. Only if there are changes does the engine consult the resource's provider for more information about the difference. This can be problematic for a number of reasons: - Not all providers do input-input comparison; some do input-state comparison - Not all providers are able to update the last deployed set of inputs when performing a refresh - Some providers--either intentionally or due to bugs--may see changes in resources whose inputs have not changed All of these situations are confusing at the very least, and the first is problematic with respect to correctness. Furthermore, the display code only renders diffs it observes rather than rendering the diffs observed by the provider, which can obscure the actual changes detected at runtime. These changes address both of these issues: - Rather than comparing the current inputs against the last inputs before calling a resource provider's Diff function, the engine calls the Diff function in all cases. - Providers may now return a list of properties that differ between the requested and actual state and the way in which they differ. This information will then be used by the CLI to render the diff appropriately. A provider may also indicate that a particular diff is between old and new inputs rather than old state and new inputs. Fixes #2453.
2019-07-01 21:34:19 +02:00
func (s *UpdateStep) Old() *resource.State { return s.old }
func (s *UpdateStep) New() *resource.State { return s.new }
func (s *UpdateStep) Res() *resource.State { return s.new }
func (s *UpdateStep) Logical() bool { return true }
func (s *UpdateStep) Diffs() []resource.PropertyKey { return s.diffs }
func (s *UpdateStep) DetailedDiff() map[string]plugin.PropertyDiff { return s.detailedDiff }
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
func (s *UpdateStep) Apply(preview bool) (resource.Status, StepCompleteFunc, error) {
// Always propagate the ID, even in previews and refreshes.
s.new.ID = s.old.ID
var resourceError error
resourceStatus := resource.StatusOK
if s.new.Custom {
// Invoke the Update RPC function for this provider:
prov, err := getProvider(s)
if err != nil {
return resource.StatusOK, nil, err
}
General prep work for refresh This change includes a bunch of refactorings I made in prep for doing refresh (first, the command, see pulumi/pulumi#1081): * The primary change is to change the way the engine's core update functionality works with respect to deploy.Source. This is the way we can plug in new sources of resource information during planning (and, soon, diffing). The way I intend to model refresh is by having a new kind of source, deploy.RefreshSource, which will let us do virtually everything about an update/diff the same way with refreshes, which avoid otherwise duplicative effort. This includes changing the planOptions (nee deployOptions) to take a new SourceFunc callback, which is responsible for creating a source specific to the kind of plan being requested. Preview, Update, and Destroy now are primarily differentiated by the kind of deploy.Source that they return, rather than sprinkling things like `if Destroying` throughout. This tidies up some logic and, more importantly, gives us precisely the refresh hook we need. * Originally, we used the deploy.NullSource for Destroy operations. This simply returns nothing, which is how Destroy works. For some reason, we were no longer doing this, and instead had some `if Destroying` cases sprinkled throughout the deploy.EvalSource. I think this is a vestige of some old way we did configuration, at least judging by a comment, which is apparently no longer relevant. * Move diff and diff-printing logic within the engine into its own pkg/engine/diff.go file, to prepare for upcoming work. * I keep noticing benign diffs anytime I regenerate protobufs. I suspect this is because we're also on different versions. I changed generate.sh to also dump the version into grpc_version.txt. At least we can understand where the diffs are coming from, decide whether to take them (i.e., a newer version), and ensure that as a team we are monotonically increasing, and not going backwards. * I also tidied up some tiny things I noticed while in there, like comments, incorrect types, lint suppressions, and so on.
2018-03-28 16:45:23 +02:00
// Update to the combination of the old "all" state, but overwritten with new inputs.
outs, rst, upderr := prov.Update(s.URN(), s.old.ID, s.old.Outputs, s.new.Inputs,
s.new.CustomTimeouts.Update, s.ignoreChanges, s.deployment.preview)
if upderr != nil {
if rst != resource.StatusPartialFailure {
return rst, nil, upderr
}
resourceError = upderr
resourceStatus = rst
if initErr, isInitErr := upderr.(*plugin.InitError); isInitErr {
s.new.InitErrors = initErr.Reasons
}
}
// Now copy any output state back in case the update triggered cascading updates to other properties.
s.new.Outputs = outs
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
Implement components This change implements core support for "components" in the Pulumi Fabric. This work is described further in pulumi/pulumi#340, where we are still discussing some of the finer points. In a nutshell, resources no longer imply external providers. It's entirely possible to have a resource that logically represents something but without having a physical manifestation that needs to be tracked and managed by our typical CRUD operations. For example, the aws/serverless/Function helper is one such type. It aggregates Lambda-related resources and exposes a nice interface. All of the Pulumi Cloud Framework resources are also examples. To indicate that a resource does participate in the usual CRUD resource provider, it simply derives from ExternalResource instead of Resource. All resources now have the ability to adopt children. This is purely a metadata/tagging thing, and will help us roll up displays, provide attribution to the developer, and even hide aspects of the resource graph as appropriate (e.g., when they are implementation details). Our use of this capability is ultra limited right now; in fact, the only place we display children is in the CLI output. For instance: + aws:serverless:Function: (create) [urn=urn:pulumi:demo::serverless::aws:serverless:Function::mylambda] => urn:pulumi:demo::serverless::aws:iam/role:Role::mylambda-iamrole => urn:pulumi:demo::serverless::aws:iam/rolePolicyAttachment:RolePolicyAttachment::mylambda-iampolicy-0 => urn:pulumi:demo::serverless::aws:lambda/function:Function::mylambda The bit indicating whether a resource is external or not is tracked in the resulting checkpoint file, along with any of its children.
2017-10-14 23:18:43 +02:00
// Finally, mark this operation as complete.
Propagate inputs to outputs during preview. (#3327) These changes restore a more-correct version of the behavior that was disabled with #3014. The original implementation of this behavior was done in the SDKs, which do not have access to the complete inputs for a resource (in particular, default values filled in by the provider during `Check` are not exposed to the SDK). This lack of information meant that the resolved output values could disagree with the typings present in a provider SDK. Exacerbating this problem was the fact that unknown values were dropped entirely, causing `undefined` values to appear in unexpected places. By doing this in the engine and allowing unknown values to be represented in a first-class manner in the SDK, we can attack both of these issues. Although this behavior is not _strictly_ consistent with respect to the resource model--in an update, a resource's output properties will come from its provider and may differ from its input properties--this behavior was present in the product for a fairly long time without significant issues. In the future, we may be able to improve the accuracy of resource outputs during a preview by allowing the provider to dry-run CRUD operations and return partially-known values where possible. These changes also introduce new APIs in the Node and Python SDKs that work with unknown values in a first-class fashion: - A new parameter to the `apply` function that indicates that the callback should be run even if the result of the apply contains unknown values - `containsUnknowns` and `isUnknown`, which return true if a value either contains nested unknown values or is exactly an unknown value - The `Unknown` type, which represents unknown values The primary use case for these APIs is to allow nested, properties with known values to be accessed via the lifted property accessor even when the containing property is not fully know. A common example of this pattern is the `metadata.name` property of a Kubernetes `Namespace` object: while other properties of the `metadata` bag may be unknown, `name` is often known. These APIs allow `ns.metadata.name` to return a known value in this case. In order to avoid exposing downlevel SDKs to unknown values--a change which could break user code by exposing it to unexpected values--a language SDK must indicate whether or not it supports first-class unknown values as part of each `RegisterResourceRequest`. These changes also allow us to avoid breaking user code with the new behavior introduced by the prior commit. Fixes #3190.
2019-11-11 21:09:34 +01:00
complete := func() { s.reg.Done(&RegisterResult{State: s.new}) }
2018-07-07 00:17:32 +02:00
if resourceError == nil {
return resourceStatus, complete, nil
2018-07-07 00:17:32 +02:00
}
return resourceStatus, complete, resourceError
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
// ReplaceStep is a logical step indicating a resource will be replaced. This is comprised of three physical steps:
// a creation of the new resource, any number of intervening updates of dependents to the new resource, and then
// a deletion of the now-replaced old resource. This logical step is primarily here for tools and visualization.
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
type ReplaceStep struct {
deployment *Deployment // the current deployment.
Defer all diffs to resource providers. (#2849) Thse changes make a subtle but critical adjustment to the process the Pulumi engine uses to determine whether or not a difference exists between a resource's actual and desired states, and adjusts the way this difference is calculated and displayed accordingly. Today, the Pulumi engine get the first chance to decide whether or not there is a difference between a resource's actual and desired states. It does this by comparing the current set of inputs for a resource (i.e. the inputs from the running Pulumi program) with the last set of inputs used to update the resource. If there is no difference between the old and new inputs, the engine decides that no change is necessary without consulting the resource's provider. Only if there are changes does the engine consult the resource's provider for more information about the difference. This can be problematic for a number of reasons: - Not all providers do input-input comparison; some do input-state comparison - Not all providers are able to update the last deployed set of inputs when performing a refresh - Some providers--either intentionally or due to bugs--may see changes in resources whose inputs have not changed All of these situations are confusing at the very least, and the first is problematic with respect to correctness. Furthermore, the display code only renders diffs it observes rather than rendering the diffs observed by the provider, which can obscure the actual changes detected at runtime. These changes address both of these issues: - Rather than comparing the current inputs against the last inputs before calling a resource provider's Diff function, the engine calls the Diff function in all cases. - Providers may now return a list of properties that differ between the requested and actual state and the way in which they differ. This information will then be used by the CLI to render the diff appropriately. A provider may also indicate that a particular diff is between old and new inputs rather than old state and new inputs. Fixes #2453.
2019-07-01 21:34:19 +02:00
old *resource.State // the state of the existing resource.
new *resource.State // the new state snapshot.
keys []resource.PropertyKey // the keys causing replacement.
diffs []resource.PropertyKey // the keys causing a diff.
detailedDiff map[string]plugin.PropertyDiff // the structured property diff.
pendingDelete bool // true if a pending deletion should happen.
}
Switch to parent pointers; display components nicely This change switches from child lists to parent pointers, in the way resource ancestries are represented. This cleans up a fair bit of the old parenting logic, including all notion of ambient parent scopes (and will notably address pulumi/pulumi#435). This lets us show a more parent/child display in the output when doing planning and updating. For instance, here is an update of a lambda's text, which is logically part of a cloud timer: * cloud:timer:Timer: (same) [urn=urn:pulumi:malta::lm-cloud::cloud:timer:Timer::lm-cts-malta-job-CleanSnapshots] * cloud:function:Function: (same) [urn=urn:pulumi:malta::lm-cloud::cloud:function:Function::lm-cts-malta-job-CleanSnapshots] * aws:serverless:Function: (same) [urn=urn:pulumi:malta::lm-cloud::aws:serverless:Function::lm-cts-malta-job-CleanSnapshots] ~ aws:lambda/function:Function: (modify) [id=lm-cts-malta-job-CleanSnapshots-fee4f3bf41280741] [urn=urn:pulumi:malta::lm-cloud::aws:lambda/function:Function::lm-cts-malta-job-CleanSnapshots] - code : archive(assets:2092f44) { // etc etc etc Note that we still get walls of text, but this will be actually quite nice when combined with pulumi/pulumi#454. I've also suppressed printing properties that didn't change during updates when --detailed was not passed, and also suppressed empty strings and zero-length arrays (since TF uses these as defaults in many places and it just makes creation and deletion quite verbose). Note that this is a far cry from everything we can possibly do here as part of pulumi/pulumi#340 (and even pulumi/pulumi#417). But it's a good start towards taming some of our output spew.
2017-11-17 03:21:41 +01:00
var _ Step = (*ReplaceStep)(nil)
func NewReplaceStep(deployment *Deployment, old, new *resource.State, keys, diffs []resource.PropertyKey,
detailedDiff map[string]plugin.PropertyDiff, pendingDelete bool) Step {
contract.Assert(old != nil)
contract.Assert(old.URN != "")
contract.Assert(old.ID != "" || !old.Custom)
contract.Assert(!old.Delete)
contract.Assert(new != nil)
contract.Assert(new.URN != "")
// contract.Assert(new.ID == "")
contract.Assert(!new.Delete)
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
return &ReplaceStep{
deployment: deployment,
old: old,
new: new,
keys: keys,
diffs: diffs,
Defer all diffs to resource providers. (#2849) Thse changes make a subtle but critical adjustment to the process the Pulumi engine uses to determine whether or not a difference exists between a resource's actual and desired states, and adjusts the way this difference is calculated and displayed accordingly. Today, the Pulumi engine get the first chance to decide whether or not there is a difference between a resource's actual and desired states. It does this by comparing the current set of inputs for a resource (i.e. the inputs from the running Pulumi program) with the last set of inputs used to update the resource. If there is no difference between the old and new inputs, the engine decides that no change is necessary without consulting the resource's provider. Only if there are changes does the engine consult the resource's provider for more information about the difference. This can be problematic for a number of reasons: - Not all providers do input-input comparison; some do input-state comparison - Not all providers are able to update the last deployed set of inputs when performing a refresh - Some providers--either intentionally or due to bugs--may see changes in resources whose inputs have not changed All of these situations are confusing at the very least, and the first is problematic with respect to correctness. Furthermore, the display code only renders diffs it observes rather than rendering the diffs observed by the provider, which can obscure the actual changes detected at runtime. These changes address both of these issues: - Rather than comparing the current inputs against the last inputs before calling a resource provider's Diff function, the engine calls the Diff function in all cases. - Providers may now return a list of properties that differ between the requested and actual state and the way in which they differ. This information will then be used by the CLI to render the diff appropriately. A provider may also indicate that a particular diff is between old and new inputs rather than old state and new inputs. Fixes #2453.
2019-07-01 21:34:19 +02:00
detailedDiff: detailedDiff,
pendingDelete: pendingDelete,
}
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
Defer all diffs to resource providers. (#2849) Thse changes make a subtle but critical adjustment to the process the Pulumi engine uses to determine whether or not a difference exists between a resource's actual and desired states, and adjusts the way this difference is calculated and displayed accordingly. Today, the Pulumi engine get the first chance to decide whether or not there is a difference between a resource's actual and desired states. It does this by comparing the current set of inputs for a resource (i.e. the inputs from the running Pulumi program) with the last set of inputs used to update the resource. If there is no difference between the old and new inputs, the engine decides that no change is necessary without consulting the resource's provider. Only if there are changes does the engine consult the resource's provider for more information about the difference. This can be problematic for a number of reasons: - Not all providers do input-input comparison; some do input-state comparison - Not all providers are able to update the last deployed set of inputs when performing a refresh - Some providers--either intentionally or due to bugs--may see changes in resources whose inputs have not changed All of these situations are confusing at the very least, and the first is problematic with respect to correctness. Furthermore, the display code only renders diffs it observes rather than rendering the diffs observed by the provider, which can obscure the actual changes detected at runtime. These changes address both of these issues: - Rather than comparing the current inputs against the last inputs before calling a resource provider's Diff function, the engine calls the Diff function in all cases. - Providers may now return a list of properties that differ between the requested and actual state and the way in which they differ. This information will then be used by the CLI to render the diff appropriately. A provider may also indicate that a particular diff is between old and new inputs rather than old state and new inputs. Fixes #2453.
2019-07-01 21:34:19 +02:00
func (s *ReplaceStep) Op() StepOp { return OpReplace }
func (s *ReplaceStep) Deployment() *Deployment { return s.deployment }
func (s *ReplaceStep) Type() tokens.Type { return s.new.Type }
Defer all diffs to resource providers. (#2849) Thse changes make a subtle but critical adjustment to the process the Pulumi engine uses to determine whether or not a difference exists between a resource's actual and desired states, and adjusts the way this difference is calculated and displayed accordingly. Today, the Pulumi engine get the first chance to decide whether or not there is a difference between a resource's actual and desired states. It does this by comparing the current set of inputs for a resource (i.e. the inputs from the running Pulumi program) with the last set of inputs used to update the resource. If there is no difference between the old and new inputs, the engine decides that no change is necessary without consulting the resource's provider. Only if there are changes does the engine consult the resource's provider for more information about the difference. This can be problematic for a number of reasons: - Not all providers do input-input comparison; some do input-state comparison - Not all providers are able to update the last deployed set of inputs when performing a refresh - Some providers--either intentionally or due to bugs--may see changes in resources whose inputs have not changed All of these situations are confusing at the very least, and the first is problematic with respect to correctness. Furthermore, the display code only renders diffs it observes rather than rendering the diffs observed by the provider, which can obscure the actual changes detected at runtime. These changes address both of these issues: - Rather than comparing the current inputs against the last inputs before calling a resource provider's Diff function, the engine calls the Diff function in all cases. - Providers may now return a list of properties that differ between the requested and actual state and the way in which they differ. This information will then be used by the CLI to render the diff appropriately. A provider may also indicate that a particular diff is between old and new inputs rather than old state and new inputs. Fixes #2453.
2019-07-01 21:34:19 +02:00
func (s *ReplaceStep) Provider() string { return s.new.Provider }
func (s *ReplaceStep) URN() resource.URN { return s.new.URN }
Defer all diffs to resource providers. (#2849) Thse changes make a subtle but critical adjustment to the process the Pulumi engine uses to determine whether or not a difference exists between a resource's actual and desired states, and adjusts the way this difference is calculated and displayed accordingly. Today, the Pulumi engine get the first chance to decide whether or not there is a difference between a resource's actual and desired states. It does this by comparing the current set of inputs for a resource (i.e. the inputs from the running Pulumi program) with the last set of inputs used to update the resource. If there is no difference between the old and new inputs, the engine decides that no change is necessary without consulting the resource's provider. Only if there are changes does the engine consult the resource's provider for more information about the difference. This can be problematic for a number of reasons: - Not all providers do input-input comparison; some do input-state comparison - Not all providers are able to update the last deployed set of inputs when performing a refresh - Some providers--either intentionally or due to bugs--may see changes in resources whose inputs have not changed All of these situations are confusing at the very least, and the first is problematic with respect to correctness. Furthermore, the display code only renders diffs it observes rather than rendering the diffs observed by the provider, which can obscure the actual changes detected at runtime. These changes address both of these issues: - Rather than comparing the current inputs against the last inputs before calling a resource provider's Diff function, the engine calls the Diff function in all cases. - Providers may now return a list of properties that differ between the requested and actual state and the way in which they differ. This information will then be used by the CLI to render the diff appropriately. A provider may also indicate that a particular diff is between old and new inputs rather than old state and new inputs. Fixes #2453.
2019-07-01 21:34:19 +02:00
func (s *ReplaceStep) Old() *resource.State { return s.old }
func (s *ReplaceStep) New() *resource.State { return s.new }
func (s *ReplaceStep) Res() *resource.State { return s.new }
func (s *ReplaceStep) Keys() []resource.PropertyKey { return s.keys }
func (s *ReplaceStep) Diffs() []resource.PropertyKey { return s.diffs }
func (s *ReplaceStep) DetailedDiff() map[string]plugin.PropertyDiff { return s.detailedDiff }
func (s *ReplaceStep) Logical() bool { return true }
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
func (s *ReplaceStep) Apply(preview bool) (resource.Status, StepCompleteFunc, error) {
// If this is a pending delete, we should have marked the old resource for deletion in the CreateReplacement step.
contract.Assert(!s.pendingDelete || s.old.Delete)
return resource.StatusOK, func() {}, nil
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
}
// ReadStep is a step indicating that an existing resources will be "read" and projected into the Pulumi object
// model. Resources that are read are marked with the "External" bit which indicates to the engine that it does
// not own this resource's lifeycle.
//
// A resource with a given URN can transition freely between an "external" state and a non-external state. If
// a URN that was previously marked "External" (i.e. was the target of a ReadStep in a previous deployment) is the
// target of a RegisterResource in the next deployment, a CreateReplacement step will be issued to indicate the
// transition from external to owned. If a URN that was previously not marked "External" is the target of a
// ReadResource in the next deployment, a ReadReplacement step will be issued to indicate the transition from owned to
// external.
type ReadStep struct {
deployment *Deployment // the deployment that produced this read
event ReadResourceEvent // the event that should be signaled upon completion
old *resource.State // the old resource state, if one exists for this urn
new *resource.State // the new resource state, to be used to query the provider
replacing bool // whether or not the new resource is replacing the old resource
}
// NewReadStep creates a new Read step.
func NewReadStep(deployment *Deployment, event ReadResourceEvent, old, new *resource.State) Step {
contract.Assert(new != nil)
contract.Assertf(new.External, "target of Read step must be marked External")
contract.Assertf(new.Custom, "target of Read step must be Custom")
// If Old was given, it's either an external resource or its ID is equal to the
// ID that we are preparing to read.
if old != nil {
contract.Assert(old.ID == new.ID || old.External)
}
return &ReadStep{
deployment: deployment,
event: event,
old: old,
new: new,
replacing: false,
}
}
// NewReadReplacementStep creates a new Read step with the `replacing` flag set. When executed,
// it will pend deletion of the "old" resource, which must not be an external resource.
func NewReadReplacementStep(deployment *Deployment, event ReadResourceEvent, old, new *resource.State) Step {
contract.Assert(new != nil)
contract.Assertf(new.External, "target of ReadReplacement step must be marked External")
contract.Assertf(new.Custom, "target of ReadReplacement step must be Custom")
contract.Assert(old != nil)
contract.Assertf(!old.External, "old target of ReadReplacement step must not be External")
return &ReadStep{
deployment: deployment,
event: event,
old: old,
new: new,
replacing: true,
}
}
func (s *ReadStep) Op() StepOp {
if s.replacing {
return OpReadReplacement
}
return OpRead
}
func (s *ReadStep) Deployment() *Deployment { return s.deployment }
func (s *ReadStep) Type() tokens.Type { return s.new.Type }
func (s *ReadStep) Provider() string { return s.new.Provider }
func (s *ReadStep) URN() resource.URN { return s.new.URN }
func (s *ReadStep) Old() *resource.State { return s.old }
func (s *ReadStep) New() *resource.State { return s.new }
func (s *ReadStep) Res() *resource.State { return s.new }
func (s *ReadStep) Logical() bool { return !s.replacing }
func (s *ReadStep) Apply(preview bool) (resource.Status, StepCompleteFunc, error) {
urn := s.new.URN
id := s.new.ID
var resourceError error
resourceStatus := resource.StatusOK
// Unlike most steps, Read steps run during previews. The only time
// we can't run is if the ID we are given is unknown.
if id == plugin.UnknownStringValue {
s.new.Outputs = resource.PropertyMap{}
} else {
prov, err := getProvider(s)
if err != nil {
return resource.StatusOK, nil, err
}
result, rst, err := prov.Read(urn, id, nil, s.new.Inputs)
if err != nil {
if rst != resource.StatusPartialFailure {
return rst, nil, err
}
resourceError = err
resourceStatus = rst
if initErr, isInitErr := err.(*plugin.InitError); isInitErr {
s.new.InitErrors = initErr.Reasons
}
}
// If there is no such resource, return an error indicating as such.
if result.Outputs == nil {
return resource.StatusOK, nil, errors.Errorf("resource '%s' does not exist", id)
}
s.new.Outputs = result.Outputs
if result.ID != "" {
s.new.ID = result.ID
}
}
// If we were asked to replace an existing, non-External resource, pend the
// deletion here.
if s.replacing {
s.old.Delete = true
}
complete := func() { s.event.Done(&ReadResult{State: s.new}) }
if resourceError == nil {
return resourceStatus, complete, nil
}
return resourceStatus, complete, resourceError
}
// RefreshStep is a step used to track the progress of a refresh operation. A refresh operation updates the an existing
// resource by reading its current state from its provider plugin. These steps are not issued by the step generator;
// instead, they are issued by the deployment executor as the optional first step in deployment execution.
type RefreshStep struct {
deployment *Deployment // the deployment that produced this refresh
old *resource.State // the old resource state, if one exists for this urn
new *resource.State // the new resource state, to be used to query the provider
done chan<- bool // the channel to use to signal completion, if any
}
// NewRefreshStep creates a new Refresh step.
func NewRefreshStep(deployment *Deployment, old *resource.State, done chan<- bool) Step {
contract.Assert(old != nil)
// NOTE: we set the new state to the old state by default so that we don't interpret step failures as deletes.
return &RefreshStep{
deployment: deployment,
old: old,
new: old,
done: done,
}
}
func (s *RefreshStep) Op() StepOp { return OpRefresh }
func (s *RefreshStep) Deployment() *Deployment { return s.deployment }
func (s *RefreshStep) Type() tokens.Type { return s.old.Type }
func (s *RefreshStep) Provider() string { return s.old.Provider }
func (s *RefreshStep) URN() resource.URN { return s.old.URN }
func (s *RefreshStep) Old() *resource.State { return s.old }
func (s *RefreshStep) New() *resource.State { return s.new }
func (s *RefreshStep) Res() *resource.State { return s.old }
func (s *RefreshStep) Logical() bool { return false }
// ResultOp returns the operation that corresponds to the change to this resource after reading its current state, if
// any.
func (s *RefreshStep) ResultOp() StepOp {
if s.new == nil {
return OpDelete
}
if s.new == s.old || s.old.Outputs.Diff(s.new.Outputs) == nil {
return OpSame
}
return OpUpdate
}
func (s *RefreshStep) Apply(preview bool) (resource.Status, StepCompleteFunc, error) {
var complete func()
if s.done != nil {
complete = func() { close(s.done) }
}
resourceID := s.old.ID
Implement more precise delete-before-replace semantics. (#2369) This implements the new algorithm for deciding which resources must be deleted due to a delete-before-replace operation. We need to compute the set of resources that may be replaced by a change to the resource under consideration. We do this by taking the complete set of transitive dependents on the resource under consideration and removing any resources that would not be replaced by changes to their dependencies. We determine whether or not a resource may be replaced by substituting unknowns for input properties that may change due to deletion of the resources their value depends on and calling the resource provider's Diff method. This is perhaps clearer when described by example. Consider the following dependency graph: A __|__ B C | _|_ D E F In this graph, all of B, C, D, E, and F transitively depend on A. It may be the case, however, that changes to the specific properties of any of those resources R that would occur if a resource on the path to A were deleted and recreated may not cause R to be replaced. For example, the edge from B to A may be a simple dependsOn edge such that a change to B does not actually influence any of B's input properties. In that case, neither B nor D would need to be deleted before A could be deleted. In order to make the above algorithm a reality, the resource monitor interface has been updated to include a map that associates an input property key with the list of resources that input property depends on. Older clients of the resource monitor will leave this map empty, in which case all input properties will be treated as depending on all dependencies of the resource. This is probably overly conservative, but it is less conservative than what we currently implement, and is certainly correct.
2019-01-28 18:46:30 +01:00
// Component, provider, and pending-replace resources never change with a refresh; just return the current state.
if !s.old.Custom || providers.IsProviderType(s.old.Type) || s.old.PendingReplacement {
return resource.StatusOK, complete, nil
}
// For a custom resource, fetch the resource's provider and read the resource's current state.
prov, err := getProvider(s)
if err != nil {
return resource.StatusOK, nil, err
}
var initErrors []string
refreshed, rst, err := prov.Read(s.old.URN, resourceID, s.old.Inputs, s.old.Outputs)
if err != nil {
if rst != resource.StatusPartialFailure {
return rst, nil, err
}
if initErr, isInitErr := err.(*plugin.InitError); isInitErr {
initErrors = initErr.Reasons
// Partial failure SHOULD NOT cause refresh to fail. Instead:
//
// 1. Warn instead that during refresh we noticed the resource has become unhealthy.
// 2. Make sure the initialization errors are persisted in the state, so that the next
// `pulumi up` will surface them to the user.
err = nil
msg := fmt.Sprintf("Refreshed resource is in an unhealthy state:\n* %s", strings.Join(initErrors, "\n* "))
s.Deployment().Diag().Warningf(diag.RawMessage(s.URN(), msg))
}
}
outputs := refreshed.Outputs
// If the provider specified new inputs for this resource, pick them up now. Otherwise, retain the current inputs.
inputs := s.old.Inputs
if refreshed.Inputs != nil {
inputs = refreshed.Inputs
}
if outputs != nil {
// There is a chance that the ID has changed. We want to allow this change to happen
// it will have changed already in the outputs, but we need to persist this change
// at a state level because the Id
if refreshed.ID != "" && refreshed.ID != resourceID {
logging.V(7).Infof("Refreshing ID; oldId=%s, newId=%s", resourceID, refreshed.ID)
resourceID = refreshed.ID
}
s.new = resource.NewState(s.old.Type, s.old.URN, s.old.Custom, s.old.Delete, resourceID, inputs, outputs,
Implement more precise delete-before-replace semantics. (#2369) This implements the new algorithm for deciding which resources must be deleted due to a delete-before-replace operation. We need to compute the set of resources that may be replaced by a change to the resource under consideration. We do this by taking the complete set of transitive dependents on the resource under consideration and removing any resources that would not be replaced by changes to their dependencies. We determine whether or not a resource may be replaced by substituting unknowns for input properties that may change due to deletion of the resources their value depends on and calling the resource provider's Diff method. This is perhaps clearer when described by example. Consider the following dependency graph: A __|__ B C | _|_ D E F In this graph, all of B, C, D, E, and F transitively depend on A. It may be the case, however, that changes to the specific properties of any of those resources R that would occur if a resource on the path to A were deleted and recreated may not cause R to be replaced. For example, the edge from B to A may be a simple dependsOn edge such that a change to B does not actually influence any of B's input properties. In that case, neither B nor D would need to be deleted before A could be deleted. In order to make the above algorithm a reality, the resource monitor interface has been updated to include a map that associates an input property key with the list of resources that input property depends on. Older clients of the resource monitor will leave this map empty, in which case all input properties will be treated as depending on all dependencies of the resource. This is probably overly conservative, but it is less conservative than what we currently implement, and is certainly correct.
2019-01-28 18:46:30 +01:00
s.old.Parent, s.old.Protect, s.old.External, s.old.Dependencies, initErrors, s.old.Provider,
2019-07-15 23:26:28 +02:00
s.old.PropertyDependencies, s.old.PendingReplacement, s.old.AdditionalSecretOutputs, s.old.Aliases,
&s.old.CustomTimeouts, s.old.ImportID)
} else {
s.new = nil
}
return rst, complete, err
}
type ImportStep struct {
deployment *Deployment // the current deployment.
reg RegisterResourceEvent // the registration intent to convey a URN back to.
original *resource.State // the original resource, if this is an import-replace.
old *resource.State // the state of the resource fetched from the provider.
new *resource.State // the newly computed state of the resource after importing.
replacing bool // true if we are replacing a Pulumi-managed resource.
planned bool // true if this import is from an import deployment.
diffs []resource.PropertyKey // any keys that differed between the user's program and the actual state.
detailedDiff map[string]plugin.PropertyDiff // the structured property diff.
ignoreChanges []string // a list of property paths to ignore when updating.
}
func NewImportStep(deployment *Deployment, reg RegisterResourceEvent, new *resource.State,
ignoreChanges []string) Step {
contract.Assert(new != nil)
contract.Assert(new.URN != "")
contract.Assert(new.ID != "")
contract.Assert(new.Custom)
contract.Assert(!new.Delete)
contract.Assert(!new.External)
return &ImportStep{
deployment: deployment,
reg: reg,
new: new,
ignoreChanges: ignoreChanges,
}
}
func NewImportReplacementStep(deployment *Deployment, reg RegisterResourceEvent, original, new *resource.State,
ignoreChanges []string) Step {
contract.Assert(original != nil)
contract.Assert(new != nil)
contract.Assert(new.URN != "")
contract.Assert(new.ID != "")
contract.Assert(new.Custom)
contract.Assert(!new.Delete)
contract.Assert(!new.External)
return &ImportStep{
deployment: deployment,
reg: reg,
original: original,
new: new,
replacing: true,
ignoreChanges: ignoreChanges,
}
}
func newImportDeploymentStep(deployment *Deployment, new *resource.State) Step {
contract.Assert(new != nil)
contract.Assert(new.URN != "")
contract.Assert(new.ID != "")
contract.Assert(new.Custom)
contract.Assert(!new.Delete)
contract.Assert(!new.External)
return &ImportStep{
deployment: deployment,
reg: noopEvent(0),
new: new,
planned: true,
}
}
func (s *ImportStep) Op() StepOp {
if s.replacing {
return OpImportReplacement
}
return OpImport
}
func (s *ImportStep) Deployment() *Deployment { return s.deployment }
func (s *ImportStep) Type() tokens.Type { return s.new.Type }
func (s *ImportStep) Provider() string { return s.new.Provider }
func (s *ImportStep) URN() resource.URN { return s.new.URN }
func (s *ImportStep) Old() *resource.State { return s.old }
func (s *ImportStep) New() *resource.State { return s.new }
func (s *ImportStep) Res() *resource.State { return s.new }
func (s *ImportStep) Logical() bool { return !s.replacing }
func (s *ImportStep) Diffs() []resource.PropertyKey { return s.diffs }
func (s *ImportStep) DetailedDiff() map[string]plugin.PropertyDiff { return s.detailedDiff }
func (s *ImportStep) Apply(preview bool) (resource.Status, StepCompleteFunc, error) {
complete := func() { s.reg.Done(&RegisterResult{State: s.new}) }
// If this is a planned import, ensure that the resource does not exist in the old state file.
if s.planned {
if _, ok := s.deployment.olds[s.new.URN]; ok {
return resource.StatusOK, nil, errors.Errorf("resource '%v' already exists", s.new.URN)
}
if s.new.Parent.Type() != resource.RootStackType {
if _, ok := s.deployment.olds[s.new.Parent]; !ok {
return resource.StatusOK, nil, errors.Errorf("unknown parent '%v' for resource '%v'",
s.new.Parent, s.new.URN)
}
}
}
// Read the current state of the resource to import. If the provider does not hand us back any inputs for the
// resource, it probably needs to be updated. If the resource does not exist at all, fail the import.
prov, err := getProvider(s)
if err != nil {
return resource.StatusOK, nil, err
}
read, rst, err := prov.Read(s.new.URN, s.new.ID, nil, nil)
if err != nil {
if initErr, isInitErr := err.(*plugin.InitError); isInitErr {
s.new.InitErrors = initErr.Reasons
} else {
return rst, nil, err
}
}
if read.Outputs == nil {
return rst, nil, errors.Errorf("resource '%v' does not exist", s.new.ID)
}
if read.Inputs == nil {
return resource.StatusOK, nil, errors.Errorf(
"provider does not support importing resources; please try updating the '%v' plugin",
s.new.URN.Type().Package())
}
if read.ID != "" {
s.new.ID = read.ID
}
s.new.Outputs = read.Outputs
// Magic up an old state so the frontend can display a proper diff. This state is the output of the just-executed
// `Read` combined with the resource identity and metadata from the desired state. This ensures that the only
// differences between the old and new states are between the inputs and outputs.
s.old = resource.NewState(s.new.Type, s.new.URN, s.new.Custom, false, s.new.ID, read.Inputs, read.Outputs,
s.new.Parent, s.new.Protect, false, s.new.Dependencies, s.new.InitErrors, s.new.Provider,
s.new.PropertyDependencies, false, nil, nil, &s.new.CustomTimeouts, s.new.ImportID)
// If this step came from an import deployment, we need to fetch any required inputs from the state.
if s.planned {
contract.Assert(len(s.new.Inputs) == 0)
pkg, err := s.deployment.schemaLoader.LoadPackage(string(s.new.Type.Package()), nil)
if err != nil {
return resource.StatusOK, nil, errors.Wrapf(err, "failed to fetch provider schema")
}
r, ok := pkg.GetResource(string(s.new.Type))
if !ok {
return resource.StatusOK, nil, errors.Errorf("unknown resource type '%v'", s.new.Type)
}
for _, p := range r.InputProperties {
if p.IsRequired {
k := resource.PropertyKey(p.Name)
s.new.Inputs[k] = s.old.Inputs[k]
}
}
}
// Check the inputs using the provider inputs for defaults.
inputs, failures, err := prov.Check(s.new.URN, s.old.Inputs, s.new.Inputs, preview)
if err != nil {
return rst, nil, err
}
if issueCheckErrors(s.deployment, s.new, s.new.URN, failures) {
return rst, nil, errors.New("one or more inputs failed to validate")
}
s.new.Inputs = inputs
// Diff the user inputs against the provider inputs. If there are any differences, fail the import unless this step
// is from an import deployment.
diff, err := diffResource(s.new.URN, s.new.ID, s.old.Inputs, s.old.Outputs, s.new.Inputs, prov, preview,
s.ignoreChanges)
if err != nil {
return rst, nil, err
}
if !s.planned {
s.diffs, s.detailedDiff = diff.ChangedKeys, diff.DetailedDiff
if diff.Changes != plugin.DiffNone {
const message = "inputs to import do not match the existing resource"
if preview {
s.deployment.ctx.Diag.Warningf(diag.StreamMessage(s.new.URN,
message+"; importing this resource will fail", 0))
} else {
err = errors.New(message)
}
}
// If we were asked to replace an existing, non-External resource, pend the deletion here.
if err == nil && s.replacing {
s.original.Delete = true
}
} else {
s.diffs, s.detailedDiff = []resource.PropertyKey{}, map[string]plugin.PropertyDiff{}
// If there were diffs between the inputs supplied by the import and the actual state of the resource, copy
// the differing properties from the actual inputs/state. This is only possible if the provider returned a
// detailed diff. If no detailed diff was returned, take the actual inputs wholesale.
if diff.Changes != plugin.DiffNone {
if diff.DetailedDiff == nil {
s.new.Inputs = s.old.Inputs
} else {
for path, pdiff := range diff.DetailedDiff {
elements, err := resource.ParsePropertyPath(path)
if err != nil {
continue
}
source := s.old.Outputs
if pdiff.InputDiff {
source = s.old.Inputs
}
if old, ok := elements.Get(resource.NewObjectProperty(source)); ok {
// Ignore failure here.
elements.Add(resource.NewObjectProperty(s.new.Inputs), old)
}
}
}
}
}
return rst, complete, err
}
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
// StepOp represents the kind of operation performed by a step. It evaluates to its string label.
type StepOp string
const (
OpSame StepOp = "same" // nothing to do.
OpCreate StepOp = "create" // creating a new resource.
OpUpdate StepOp = "update" // updating an existing resource.
OpDelete StepOp = "delete" // deleting an existing resource.
OpReplace StepOp = "replace" // replacing a resource with a new one.
OpCreateReplacement StepOp = "create-replacement" // creating a new resource for a replacement.
OpDeleteReplaced StepOp = "delete-replaced" // deleting an existing resource after replacement.
OpRead StepOp = "read" // reading an existing resource.
OpReadReplacement StepOp = "read-replacement" // reading an existing resource for a replacement.
OpRefresh StepOp = "refresh" // refreshing an existing resource.
OpReadDiscard StepOp = "discard" // removing a resource that was read.
OpDiscardReplaced StepOp = "discard-replaced" // discarding a read resource that was replaced.
Implement more precise delete-before-replace semantics. (#2369) This implements the new algorithm for deciding which resources must be deleted due to a delete-before-replace operation. We need to compute the set of resources that may be replaced by a change to the resource under consideration. We do this by taking the complete set of transitive dependents on the resource under consideration and removing any resources that would not be replaced by changes to their dependencies. We determine whether or not a resource may be replaced by substituting unknowns for input properties that may change due to deletion of the resources their value depends on and calling the resource provider's Diff method. This is perhaps clearer when described by example. Consider the following dependency graph: A __|__ B C | _|_ D E F In this graph, all of B, C, D, E, and F transitively depend on A. It may be the case, however, that changes to the specific properties of any of those resources R that would occur if a resource on the path to A were deleted and recreated may not cause R to be replaced. For example, the edge from B to A may be a simple dependsOn edge such that a change to B does not actually influence any of B's input properties. In that case, neither B nor D would need to be deleted before A could be deleted. In order to make the above algorithm a reality, the resource monitor interface has been updated to include a map that associates an input property key with the list of resources that input property depends on. Older clients of the resource monitor will leave this map empty, in which case all input properties will be treated as depending on all dependencies of the resource. This is probably overly conservative, but it is less conservative than what we currently implement, and is certainly correct.
2019-01-28 18:46:30 +01:00
OpRemovePendingReplace StepOp = "remove-pending-replace" // removing a pending replace resource.
OpImport StepOp = "import" // import an existing resource.
OpImportReplacement StepOp = "import-replacement" // replace an existing resource with an imported resource.
)
// StepOps contains the full set of step operation types.
var StepOps = []StepOp{
OpSame,
OpCreate,
OpUpdate,
OpDelete,
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
OpReplace,
OpCreateReplacement,
OpDeleteReplaced,
OpRead,
OpReadReplacement,
OpRefresh,
OpReadDiscard,
OpDiscardReplaced,
Implement more precise delete-before-replace semantics. (#2369) This implements the new algorithm for deciding which resources must be deleted due to a delete-before-replace operation. We need to compute the set of resources that may be replaced by a change to the resource under consideration. We do this by taking the complete set of transitive dependents on the resource under consideration and removing any resources that would not be replaced by changes to their dependencies. We determine whether or not a resource may be replaced by substituting unknowns for input properties that may change due to deletion of the resources their value depends on and calling the resource provider's Diff method. This is perhaps clearer when described by example. Consider the following dependency graph: A __|__ B C | _|_ D E F In this graph, all of B, C, D, E, and F transitively depend on A. It may be the case, however, that changes to the specific properties of any of those resources R that would occur if a resource on the path to A were deleted and recreated may not cause R to be replaced. For example, the edge from B to A may be a simple dependsOn edge such that a change to B does not actually influence any of B's input properties. In that case, neither B nor D would need to be deleted before A could be deleted. In order to make the above algorithm a reality, the resource monitor interface has been updated to include a map that associates an input property key with the list of resources that input property depends on. Older clients of the resource monitor will leave this map empty, in which case all input properties will be treated as depending on all dependencies of the resource. This is probably overly conservative, but it is less conservative than what we currently implement, and is certainly correct.
2019-01-28 18:46:30 +01:00
OpRemovePendingReplace,
OpImport,
OpImportReplacement,
}
// Color returns a suggested color for lines of this op type.
func (op StepOp) Color() string {
switch op {
case OpSame:
Switch to parent pointers; display components nicely This change switches from child lists to parent pointers, in the way resource ancestries are represented. This cleans up a fair bit of the old parenting logic, including all notion of ambient parent scopes (and will notably address pulumi/pulumi#435). This lets us show a more parent/child display in the output when doing planning and updating. For instance, here is an update of a lambda's text, which is logically part of a cloud timer: * cloud:timer:Timer: (same) [urn=urn:pulumi:malta::lm-cloud::cloud:timer:Timer::lm-cts-malta-job-CleanSnapshots] * cloud:function:Function: (same) [urn=urn:pulumi:malta::lm-cloud::cloud:function:Function::lm-cts-malta-job-CleanSnapshots] * aws:serverless:Function: (same) [urn=urn:pulumi:malta::lm-cloud::aws:serverless:Function::lm-cts-malta-job-CleanSnapshots] ~ aws:lambda/function:Function: (modify) [id=lm-cts-malta-job-CleanSnapshots-fee4f3bf41280741] [urn=urn:pulumi:malta::lm-cloud::aws:lambda/function:Function::lm-cts-malta-job-CleanSnapshots] - code : archive(assets:2092f44) { // etc etc etc Note that we still get walls of text, but this will be actually quite nice when combined with pulumi/pulumi#454. I've also suppressed printing properties that didn't change during updates when --detailed was not passed, and also suppressed empty strings and zero-length arrays (since TF uses these as defaults in many places and it just makes creation and deletion quite verbose). Note that this is a far cry from everything we can possibly do here as part of pulumi/pulumi#340 (and even pulumi/pulumi#417). But it's a good start towards taming some of our output spew.
2017-11-17 03:21:41 +01:00
return colors.SpecUnimportant
case OpCreate, OpImport:
return colors.SpecCreate
case OpDelete:
return colors.SpecDelete
case OpUpdate:
return colors.SpecUpdate
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
case OpReplace:
return colors.SpecReplace
case OpCreateReplacement:
return colors.SpecCreateReplacement
case OpDeleteReplaced:
return colors.SpecDeleteReplaced
case OpRead:
return colors.SpecRead
case OpReadReplacement, OpImportReplacement:
return colors.SpecReplace
case OpRefresh:
return colors.SpecUpdate
case OpReadDiscard, OpDiscardReplaced:
return colors.SpecDelete
default:
contract.Failf("Unrecognized resource step op: '%v'", op)
return ""
}
}
// Prefix returns a suggested prefix for lines of this op type.
func (op StepOp) Prefix() string {
return op.Color() + op.RawPrefix()
}
// RawPrefix returns the uncolorized prefix text.
func (op StepOp) RawPrefix() string {
switch op {
case OpSame:
return " "
case OpCreate:
return "+ "
case OpDelete:
return "- "
case OpUpdate:
return "~ "
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
case OpReplace:
return "+-"
case OpCreateReplacement:
return "++"
case OpDeleteReplaced:
return "--"
case OpRead:
return "> "
case OpReadReplacement:
return ">>"
case OpRefresh:
return "~ "
case OpReadDiscard:
return "< "
case OpDiscardReplaced:
return "<<"
case OpImport:
return "= "
case OpImportReplacement:
return "=>"
default:
contract.Failf("Unrecognized resource step op: %v", op)
return ""
}
}
func (op StepOp) PastTense() string {
switch op {
case OpSame, OpCreate, OpDelete, OpReplace, OpCreateReplacement, OpDeleteReplaced, OpUpdate, OpReadReplacement:
return string(op) + "d"
case OpRefresh:
return "refreshed"
case OpRead:
return "read"
case OpReadDiscard, OpDiscardReplaced:
return "discarded"
case OpImport, OpImportReplacement:
return "imported"
default:
2017-11-29 19:06:51 +01:00
contract.Failf("Unexpected resource step op: %v", op)
return ""
}
}
// Suffix returns a suggested suffix for lines of this op type.
func (op StepOp) Suffix() string {
switch op {
case OpCreateReplacement, OpUpdate, OpReplace, OpReadReplacement, OpRefresh, OpImportReplacement:
Implement `get` functions on all resources This change implements the `get` function for resources. Per pulumi/lumi#83, this allows Lumi scripts to actually read from the target environment. For example, we can now look up a SecurityGroup from its ARN: let group = aws.ec2.SecurityGroup.get( "arn:aws:ec2:us-west-2:153052954103:security-group:sg-02150d79"); The returned object is a fully functional resource object. So, we can then link it up with an EC2 instance, for example, in the usual ways: let instance = new aws.ec2.Instance(..., { securityGroups: [ group ], }); This didn't require any changes to the RPC or provider model, since we already implement the Get function. There are a few loose ends; two are short term: 1) URNs are not rehydrated. 2) Query is not yet implemented. One is mid-term: 3) We probably want a URN-based lookup function. But we will likely wait until we tackle pulumi/lumi#109 before adding this. And one is long term (and subtle): 4) These amount to I/O and are not repeatable! A change in the target environment may cause a script to generate a different plan intermittently. Most likely we want to apply a different kind of deployment "policy" for such scripts. These are inching towards the scripting model of pulumi/lumi#121, which is an entirely different beast than the repeatable immutable infrastructure deployments. Finally, it is worth noting that with this, we have some of the fundamental underpinnings required to finally tackle "inference" (pulumi/lumi#142).
2017-06-20 02:24:00 +02:00
return colors.Reset // updates and replacements colorize individual lines; get has none
}
return ""
}
// getProvider fetches the provider for the given step.
func getProvider(s Step) (plugin.Provider, error) {
if providers.IsProviderType(s.Type()) {
return s.Deployment().providers, nil
Implement first-class providers. (#1695) ### First-Class Providers These changes implement support for first-class providers. First-class providers are provider plugins that are exposed as resources via the Pulumi programming model so that they may be explicitly and multiply instantiated. Each instance of a provider resource may be configured differently, and configuration parameters may be source from the outputs of other resources. ### Provider Plugin Changes In order to accommodate the need to verify and diff provider configuration and configure providers without complete configuration information, these changes adjust the high-level provider plugin interface. Two new methods for validating a provider's configuration and diffing changes to the same have been added (`CheckConfig` and `DiffConfig`, respectively), and the type of the configuration bag accepted by `Configure` has been changed to a `PropertyMap`. These changes have not yet been reflected in the provider plugin gRPC interface. We will do this in a set of follow-up changes. Until then, these methods are implemented by adapters: - `CheckConfig` validates that all configuration parameters are string or unknown properties. This is necessary because existing plugins only accept string-typed configuration values. - `DiffConfig` either returns "never replace" if all configuration values are known or "must replace" if any configuration value is unknown. The justification for this behavior is given [here](https://github.com/pulumi/pulumi/pull/1695/files#diff-a6cd5c7f337665f5bb22e92ca5f07537R106) - `Configure` converts the config bag to a legacy config map and configures the provider plugin if all config values are known. If any config value is unknown, the underlying plugin is not configured and the provider may only perform `Check`, `Read`, and `Invoke`, all of which return empty results. We justify this behavior becuase it is only possible during a preview and provides the best experience we can manage with the existing gRPC interface. ### Resource Model Changes Providers are now exposed as resources that participate in a stack's dependency graph. Like other resources, they are explicitly created, may have multiple instances, and may have dependencies on other resources. Providers are referred to using provider references, which are a combination of the provider's URN and its ID. This design addresses the need during a preview to refer to providers that have not yet been physically created and therefore have no ID. All custom resources that are not themselves providers must specify a single provider via a provider reference. The named provider will be used to manage that resource's CRUD operations. If a resource's provider reference changes, the resource must be replaced. Though its URN is not present in the resource's dependency list, the provider should be treated as a dependency of the resource when topologically sorting the dependency graph. Finally, `Invoke` operations must now specify a provider to use for the invocation via a provider reference. ### Engine Changes First-class providers support requires a few changes to the engine: - The engine must have some way to map from provider references to provider plugins. It must be possible to add providers from a stack's checkpoint to this map and to register new/updated providers during the execution of a plan in response to CRUD operations on provider resources. - In order to support updating existing stacks using existing Pulumi programs that may not explicitly instantiate providers, the engine must be able to manage the "default" providers for each package referenced by a checkpoint or Pulumi program. The configuration for a "default" provider is taken from the stack's configuration data. The former need is addressed by adding a provider registry type that is responsible for managing all of the plugins required by a plan. In addition to loading plugins froma checkpoint and providing the ability to map from a provider reference to a provider plugin, this type serves as the provider plugin for providers themselves (i.e. it is the "provider provider"). The latter need is solved via two relatively self-contained changes to plan setup and the eval source. During plan setup, the old checkpoint is scanned for custom resources that do not have a provider reference in order to compute the set of packages that require a default provider. Once this set has been computed, the required default provider definitions are conjured and prepended to the checkpoint's resource list. Each resource that requires a default provider is then updated to refer to the default provider for its package. While an eval source is running, each custom resource registration, resource read, and invoke that does not name a provider is trapped before being returned by the source iterator. If no default provider for the appropriate package has been registered, the eval source synthesizes an appropriate registration, waits for it to complete, and records the registered provider's reference. This reference is injected into the original request, which is then processed as usual. If a default provider was already registered, the recorded reference is used and no new registration occurs. ### SDK Changes These changes only expose first-class providers from the Node.JS SDK. - A new abstract class, `ProviderResource`, can be subclassed and used to instantiate first-class providers. - A new field in `ResourceOptions`, `provider`, can be used to supply a particular provider instance to manage a `CustomResource`'s CRUD operations. - A new type, `InvokeOptions`, can be used to specify options that control the behavior of a call to `pulumi.runtime.invoke`. This type includes a `provider` field that is analogous to `ResourceOptions.provider`.
2018-08-07 02:50:29 +02:00
}
ref, err := providers.ParseReference(s.Provider())
if err != nil {
return nil, errors.Errorf("bad provider reference '%v' for resource %v: %v", s.Provider(), s.URN(), err)
Implement first-class providers. (#1695) ### First-Class Providers These changes implement support for first-class providers. First-class providers are provider plugins that are exposed as resources via the Pulumi programming model so that they may be explicitly and multiply instantiated. Each instance of a provider resource may be configured differently, and configuration parameters may be source from the outputs of other resources. ### Provider Plugin Changes In order to accommodate the need to verify and diff provider configuration and configure providers without complete configuration information, these changes adjust the high-level provider plugin interface. Two new methods for validating a provider's configuration and diffing changes to the same have been added (`CheckConfig` and `DiffConfig`, respectively), and the type of the configuration bag accepted by `Configure` has been changed to a `PropertyMap`. These changes have not yet been reflected in the provider plugin gRPC interface. We will do this in a set of follow-up changes. Until then, these methods are implemented by adapters: - `CheckConfig` validates that all configuration parameters are string or unknown properties. This is necessary because existing plugins only accept string-typed configuration values. - `DiffConfig` either returns "never replace" if all configuration values are known or "must replace" if any configuration value is unknown. The justification for this behavior is given [here](https://github.com/pulumi/pulumi/pull/1695/files#diff-a6cd5c7f337665f5bb22e92ca5f07537R106) - `Configure` converts the config bag to a legacy config map and configures the provider plugin if all config values are known. If any config value is unknown, the underlying plugin is not configured and the provider may only perform `Check`, `Read`, and `Invoke`, all of which return empty results. We justify this behavior becuase it is only possible during a preview and provides the best experience we can manage with the existing gRPC interface. ### Resource Model Changes Providers are now exposed as resources that participate in a stack's dependency graph. Like other resources, they are explicitly created, may have multiple instances, and may have dependencies on other resources. Providers are referred to using provider references, which are a combination of the provider's URN and its ID. This design addresses the need during a preview to refer to providers that have not yet been physically created and therefore have no ID. All custom resources that are not themselves providers must specify a single provider via a provider reference. The named provider will be used to manage that resource's CRUD operations. If a resource's provider reference changes, the resource must be replaced. Though its URN is not present in the resource's dependency list, the provider should be treated as a dependency of the resource when topologically sorting the dependency graph. Finally, `Invoke` operations must now specify a provider to use for the invocation via a provider reference. ### Engine Changes First-class providers support requires a few changes to the engine: - The engine must have some way to map from provider references to provider plugins. It must be possible to add providers from a stack's checkpoint to this map and to register new/updated providers during the execution of a plan in response to CRUD operations on provider resources. - In order to support updating existing stacks using existing Pulumi programs that may not explicitly instantiate providers, the engine must be able to manage the "default" providers for each package referenced by a checkpoint or Pulumi program. The configuration for a "default" provider is taken from the stack's configuration data. The former need is addressed by adding a provider registry type that is responsible for managing all of the plugins required by a plan. In addition to loading plugins froma checkpoint and providing the ability to map from a provider reference to a provider plugin, this type serves as the provider plugin for providers themselves (i.e. it is the "provider provider"). The latter need is solved via two relatively self-contained changes to plan setup and the eval source. During plan setup, the old checkpoint is scanned for custom resources that do not have a provider reference in order to compute the set of packages that require a default provider. Once this set has been computed, the required default provider definitions are conjured and prepended to the checkpoint's resource list. Each resource that requires a default provider is then updated to refer to the default provider for its package. While an eval source is running, each custom resource registration, resource read, and invoke that does not name a provider is trapped before being returned by the source iterator. If no default provider for the appropriate package has been registered, the eval source synthesizes an appropriate registration, waits for it to complete, and records the registered provider's reference. This reference is injected into the original request, which is then processed as usual. If a default provider was already registered, the recorded reference is used and no new registration occurs. ### SDK Changes These changes only expose first-class providers from the Node.JS SDK. - A new abstract class, `ProviderResource`, can be subclassed and used to instantiate first-class providers. - A new field in `ResourceOptions`, `provider`, can be used to supply a particular provider instance to manage a `CustomResource`'s CRUD operations. - A new type, `InvokeOptions`, can be used to specify options that control the behavior of a call to `pulumi.runtime.invoke`. This type includes a `provider` field that is analogous to `ResourceOptions.provider`.
2018-08-07 02:50:29 +02:00
}
provider, ok := s.Deployment().GetProvider(ref)
Implement first-class providers. (#1695) ### First-Class Providers These changes implement support for first-class providers. First-class providers are provider plugins that are exposed as resources via the Pulumi programming model so that they may be explicitly and multiply instantiated. Each instance of a provider resource may be configured differently, and configuration parameters may be source from the outputs of other resources. ### Provider Plugin Changes In order to accommodate the need to verify and diff provider configuration and configure providers without complete configuration information, these changes adjust the high-level provider plugin interface. Two new methods for validating a provider's configuration and diffing changes to the same have been added (`CheckConfig` and `DiffConfig`, respectively), and the type of the configuration bag accepted by `Configure` has been changed to a `PropertyMap`. These changes have not yet been reflected in the provider plugin gRPC interface. We will do this in a set of follow-up changes. Until then, these methods are implemented by adapters: - `CheckConfig` validates that all configuration parameters are string or unknown properties. This is necessary because existing plugins only accept string-typed configuration values. - `DiffConfig` either returns "never replace" if all configuration values are known or "must replace" if any configuration value is unknown. The justification for this behavior is given [here](https://github.com/pulumi/pulumi/pull/1695/files#diff-a6cd5c7f337665f5bb22e92ca5f07537R106) - `Configure` converts the config bag to a legacy config map and configures the provider plugin if all config values are known. If any config value is unknown, the underlying plugin is not configured and the provider may only perform `Check`, `Read`, and `Invoke`, all of which return empty results. We justify this behavior becuase it is only possible during a preview and provides the best experience we can manage with the existing gRPC interface. ### Resource Model Changes Providers are now exposed as resources that participate in a stack's dependency graph. Like other resources, they are explicitly created, may have multiple instances, and may have dependencies on other resources. Providers are referred to using provider references, which are a combination of the provider's URN and its ID. This design addresses the need during a preview to refer to providers that have not yet been physically created and therefore have no ID. All custom resources that are not themselves providers must specify a single provider via a provider reference. The named provider will be used to manage that resource's CRUD operations. If a resource's provider reference changes, the resource must be replaced. Though its URN is not present in the resource's dependency list, the provider should be treated as a dependency of the resource when topologically sorting the dependency graph. Finally, `Invoke` operations must now specify a provider to use for the invocation via a provider reference. ### Engine Changes First-class providers support requires a few changes to the engine: - The engine must have some way to map from provider references to provider plugins. It must be possible to add providers from a stack's checkpoint to this map and to register new/updated providers during the execution of a plan in response to CRUD operations on provider resources. - In order to support updating existing stacks using existing Pulumi programs that may not explicitly instantiate providers, the engine must be able to manage the "default" providers for each package referenced by a checkpoint or Pulumi program. The configuration for a "default" provider is taken from the stack's configuration data. The former need is addressed by adding a provider registry type that is responsible for managing all of the plugins required by a plan. In addition to loading plugins froma checkpoint and providing the ability to map from a provider reference to a provider plugin, this type serves as the provider plugin for providers themselves (i.e. it is the "provider provider"). The latter need is solved via two relatively self-contained changes to plan setup and the eval source. During plan setup, the old checkpoint is scanned for custom resources that do not have a provider reference in order to compute the set of packages that require a default provider. Once this set has been computed, the required default provider definitions are conjured and prepended to the checkpoint's resource list. Each resource that requires a default provider is then updated to refer to the default provider for its package. While an eval source is running, each custom resource registration, resource read, and invoke that does not name a provider is trapped before being returned by the source iterator. If no default provider for the appropriate package has been registered, the eval source synthesizes an appropriate registration, waits for it to complete, and records the registered provider's reference. This reference is injected into the original request, which is then processed as usual. If a default provider was already registered, the recorded reference is used and no new registration occurs. ### SDK Changes These changes only expose first-class providers from the Node.JS SDK. - A new abstract class, `ProviderResource`, can be subclassed and used to instantiate first-class providers. - A new field in `ResourceOptions`, `provider`, can be used to supply a particular provider instance to manage a `CustomResource`'s CRUD operations. - A new type, `InvokeOptions`, can be used to specify options that control the behavior of a call to `pulumi.runtime.invoke`. This type includes a `provider` field that is analogous to `ResourceOptions.provider`.
2018-08-07 02:50:29 +02:00
if !ok {
return nil, errors.Errorf("unknown provider '%v' for resource %v", s.Provider(), s.URN())
Implement first-class providers. (#1695) ### First-Class Providers These changes implement support for first-class providers. First-class providers are provider plugins that are exposed as resources via the Pulumi programming model so that they may be explicitly and multiply instantiated. Each instance of a provider resource may be configured differently, and configuration parameters may be source from the outputs of other resources. ### Provider Plugin Changes In order to accommodate the need to verify and diff provider configuration and configure providers without complete configuration information, these changes adjust the high-level provider plugin interface. Two new methods for validating a provider's configuration and diffing changes to the same have been added (`CheckConfig` and `DiffConfig`, respectively), and the type of the configuration bag accepted by `Configure` has been changed to a `PropertyMap`. These changes have not yet been reflected in the provider plugin gRPC interface. We will do this in a set of follow-up changes. Until then, these methods are implemented by adapters: - `CheckConfig` validates that all configuration parameters are string or unknown properties. This is necessary because existing plugins only accept string-typed configuration values. - `DiffConfig` either returns "never replace" if all configuration values are known or "must replace" if any configuration value is unknown. The justification for this behavior is given [here](https://github.com/pulumi/pulumi/pull/1695/files#diff-a6cd5c7f337665f5bb22e92ca5f07537R106) - `Configure` converts the config bag to a legacy config map and configures the provider plugin if all config values are known. If any config value is unknown, the underlying plugin is not configured and the provider may only perform `Check`, `Read`, and `Invoke`, all of which return empty results. We justify this behavior becuase it is only possible during a preview and provides the best experience we can manage with the existing gRPC interface. ### Resource Model Changes Providers are now exposed as resources that participate in a stack's dependency graph. Like other resources, they are explicitly created, may have multiple instances, and may have dependencies on other resources. Providers are referred to using provider references, which are a combination of the provider's URN and its ID. This design addresses the need during a preview to refer to providers that have not yet been physically created and therefore have no ID. All custom resources that are not themselves providers must specify a single provider via a provider reference. The named provider will be used to manage that resource's CRUD operations. If a resource's provider reference changes, the resource must be replaced. Though its URN is not present in the resource's dependency list, the provider should be treated as a dependency of the resource when topologically sorting the dependency graph. Finally, `Invoke` operations must now specify a provider to use for the invocation via a provider reference. ### Engine Changes First-class providers support requires a few changes to the engine: - The engine must have some way to map from provider references to provider plugins. It must be possible to add providers from a stack's checkpoint to this map and to register new/updated providers during the execution of a plan in response to CRUD operations on provider resources. - In order to support updating existing stacks using existing Pulumi programs that may not explicitly instantiate providers, the engine must be able to manage the "default" providers for each package referenced by a checkpoint or Pulumi program. The configuration for a "default" provider is taken from the stack's configuration data. The former need is addressed by adding a provider registry type that is responsible for managing all of the plugins required by a plan. In addition to loading plugins froma checkpoint and providing the ability to map from a provider reference to a provider plugin, this type serves as the provider plugin for providers themselves (i.e. it is the "provider provider"). The latter need is solved via two relatively self-contained changes to plan setup and the eval source. During plan setup, the old checkpoint is scanned for custom resources that do not have a provider reference in order to compute the set of packages that require a default provider. Once this set has been computed, the required default provider definitions are conjured and prepended to the checkpoint's resource list. Each resource that requires a default provider is then updated to refer to the default provider for its package. While an eval source is running, each custom resource registration, resource read, and invoke that does not name a provider is trapped before being returned by the source iterator. If no default provider for the appropriate package has been registered, the eval source synthesizes an appropriate registration, waits for it to complete, and records the registered provider's reference. This reference is injected into the original request, which is then processed as usual. If a default provider was already registered, the recorded reference is used and no new registration occurs. ### SDK Changes These changes only expose first-class providers from the Node.JS SDK. - A new abstract class, `ProviderResource`, can be subclassed and used to instantiate first-class providers. - A new field in `ResourceOptions`, `provider`, can be used to supply a particular provider instance to manage a `CustomResource`'s CRUD operations. - A new type, `InvokeOptions`, can be used to specify options that control the behavior of a call to `pulumi.runtime.invoke`. This type includes a `provider` field that is analogous to `ResourceOptions.provider`.
2018-08-07 02:50:29 +02:00
}
return provider, nil
}