Hit this while trying to add some validation checks to runs of Preview while using plans. Seems one test actually was assuming this was the case already and just hasn't been running it's expected validate function.
* Update command respects `--target-dependents`
* Update CHANGELOG_PENDING.md
* Depend on parent and provider
* Add tests for new feature
* Separate predicate and mutation in code
* Remove `targetDependentsForUpdate`
* Refactor `isTargetedForUpdate`
* Add very important nil check
* Fix issue with --target deletion dependant calculation
The code that computed --target deletion dependants was not correct.
It used parent/child component relationships, but did not respect actual
DAG dependencies. As a result, it could erroneously leave hanging
references to resources that no longer exist after performing a
`pulumi destroy --target X` operation. This manifested in bugs like
https://github.com/pulumi/pulumi/issues/6283, which is fixed by this
change. The solution is to compute the (transitive!) dependency graph
correctly, factoring in both parent/child, as well as explicit and
implicit, dependencies. The existing logic does the correct thing once
we do this. I've also added tests for this area, including regression
tests that cover transitive dependency relationships, as well as ones
that would cause an infinite loop given a naive implementation.
* Add a changelog entry
* Fix failing test to include all destroyed targets
Unless I'm missing something, the entire tree should be deleted
in this test case because A is the ancestor for the entire tree.
* Use DependencyGraph to compute dependents
Per code review feedback from @pgavlin.
Co-authored-by: Anton Tayanovskyy <anton@pulumi.com>
A few identically-typed variables got confused with the changes in #7737.
The confusion caused empty property values to be included in resources
that had any dependencies on other resources, which confused the
unmarshaling code for Go multi-language components. These changes fix
the typo and restore the original behavior, which is to omit empty
property values.
Co-authored-by: Emiliza Gutierrez <emiliza@pulumi.com>
Do not return the inputs as the state for update previews that use an
unconfigured provider. Returning the inputs as the state allows the
language SDKs to incorrectly treat unknown properties as known (because
we can't call `Update` on an unconfigured provider, we can't know which
properties are unknown). Users can re-enable the existing behavior by
setting the `PULUMI_LEGACY_PROVIDER_PREVIEW` environment variable to a
truthy value (e.g. `1`, `true`, etc.).
Most users will be unaffected by these changes. The most common programs
that may be affected are those that combine the creation of a managed
Kubernetes cluster with the deployment of applications to that cluster. These
programs generally need to configure a k8s provider instance by constructing
a kubeconfig from the output of the managed k8s cluster. Any changes to the
cluster that cause the kubeconfig to be unknown then cause the provider to
go unconfigured at runtime. Prior to these changes, resources managed by the
k8s provider would have some known outputs in this scenario, as the engine
would treat the resource's input values as its output values. After these changes,
the resource's outputs will be treated as unknown. The most frequent affect
that this has is that applies/stack outputs that depend on the outputs of
a k8s resource managed by a provider with an unknown kubeconfig will not
run/be displayed as `output`s during previews, respectively.
We might be able to improve on this by taking advantage of schema
information and filling in unknown values for properties that do not
exist in the inputs.
Fixes#7521.
Co-authored-by: Justin Van Patten <jvp@justinvp.com>
Co-authored-by: Luke Hoban <luke@pulumi.com>
Adds a new resource option to force replacement when certain properties report changes, even if the resource provider itself does not require a replacement.
Fixes#6753.
Co-authored-by: Levi Blackstone <levi@pulumi.com>
* Send plugin install output to stderr
We currently send plugin install output to stdout. This interferes
with --json (#5747), automation API scenarios, and in general is bad
CLI hygiene. This change sends plugin output to stdout instead.
* Add a changelog entry
* Propagate workspace.Project metadata to plugin init
* Get to a working fix
* Propagate Root via plugin context
* Propagate root instead of yaml path
* Revert out unnecessary parameter propagation
* Root is now always absolute at this point; simplify code and docs
* Drop python conditional and propagate unused -root to all lang hosts
* Add tests that fail before and pass after
* Lint
* Add changelog entry
Fixes: #6934
With this snippet of code:
```
func main() {
pulumi.Run(func(ctx *pulumi.Context) error {
// Create an AWS resource (S3 Bucket)
vpc, err := ec2.NewVpc(ctx, "main", &ec2.VpcArgs{
CidrBlock: pulumi.String("10.0.0.0/16"),
})
if err != nil {
return err
}
rt, err := ec2.NewRouteTable(ctx, "example", &ec2.RouteTableArgs{
VpcId: vpc.ID(),
})
// Export the name of the bucket
ctx.Export("rt", rt)
return nil
})
}
```
the CLI would panic on the diff as follows:
```
panic: fatal: An assertion has failed
goroutine 249 [running]:
github.com/pulumi/pulumi/sdk/v3/go/common/util/contract.failfast(...)
/private/tmp/pulumi-20210422-70582-1bpvlru/sdk/go/common/util/contract/failfast.go:23
github.com/pulumi/pulumi/sdk/v3/go/common/util/contract.Assert(...)
/private/tmp/pulumi-20210422-70582-1bpvlru/sdk/go/common/util/contract/assert.go:26
github.com/pulumi/pulumi/pkg/v3/engine.printPropertyValue(0xc0005d41b0, 0x57cce00, 0xc001da9050, 0x0, 0x1, 0x5932853, 0x4, 0x0)
/private/tmp/pulumi-20210422-70582-1bpvlru/pkg/engine/diff.go:511 +0x1485
```
This was due to the entire object being added to the output and
the property being a ResourceReference
On the changing of the code to use a switch statement, we can
now include the ResourceReference and ensure that we catch any
missing case statements with a panic as default
This means the same piece of code now outputs to the CLI as
follows:
```
Outputs:
rt: {
URN: "urn:pulumi:dev::testing-new-engine-diff::aws:ec2/routeTable:RouteTable::example"
ID : "rtb-09b37608ec34f3b49"
PackageVersion: ""
}
Resources:
3 unchanged
Duration: 2s
```
* Fix resource-ref-as-ID marshaling. (#6125)
This reapplies 2f0dba23ab.
* Fix malformed resource value bug
PR #6125 introduced a bug by marshaling resource
ids as PropertyValues, but not handling that case on
the unmarshaling side. The previous code assumed
that the id was a simple string value. This bug prevents
any stack update operations (preview, update, destroy,
refresh). Since this change was already
released, we must now handle both cases in the
unmarshaling code.
* Add resource ref unit tests for the Go SDK. (#6142)
This reapplies 3d505912b8.
Co-authored-by: Pat Gavlin <pat@pulumi.com>
When marshaling a resource reference as its ID (i.e. when
opts.KeepResources is false, as it will be in the case of downlevel SDKs
and resource providers), we must take care to marshal/unmarshal an empty
ID as the unknown property value.
This includes the following changes to the resource ref APIs:
- Bifurcate resource reference creation into two methods: one for
creating references to custom resources and one for creating
references to component resources.
- Store the ID in a resource reference as a PropertyValue s.t. it can be
computed.
- Add a helper method for retrieving the ID as a string + an indicator of
whether or not the reference has an ID.
Fixes#5939.
Although raw provider resources accept an input that allows a user to
specifiy a provider version to use, this input is not reflected in
current SDK code generation. Furthermore, we already have a method to
specify the provider version that should be used for a resource: the
"version" resource option. These changes update the code that handles
provider resource registrations to autmoatically populate the version
input from the "version" resource option if the option is present.
Fixes https://github.com/pulumi/pulumi-azure/issues/803.
These changes are a combination of three commits, each of which
contributes to the testing and/or fixing of a problem with marshaling
unknowns in `plugin.provider.Update` when `preview` is true.
## deploytest: add support for gRPC adapters.
These changes add support for communicating with providers using the
gRPC adapters to the deploytest pacakage. This makes it easier to test
the gRPC adapters across typical lifecycle patterns.
Supporting these changes are two additions to the `resource/plugin`
package:
1. A type that bridges between the `plugin.Provider` interface and the
`pulumirpc.ResourceProviderServer`
2. A function to create a `plugin.Provider` given a
`pulumirpc.ResourceProviderClient`
The deploytest package uses these to wrap an in-process
`plugin.Provider` in a gRPC interface and connect to it without using
the default plugin host, respectively.
## pulumi_test: test provider preview over gRPC.
Add a test that runs the provider preview lifecycle, but using a
provider that communicates over gRPC.
## gRPC bridge: fix unknowns in `Update` previews
Set the `KeepUnknowns` and `RejectUnknowns` bits in the `MarshalOptions`
used when unmarshaling update results to preserve unknowns during a
preview and reject them otherwise.
These changes also set the `RejectUnknowns` bit in the `MarshalOptions`
used by `Create` if `preview` is false, and fix a bug in the array
unmarshaler that could cause out-of-bounds accesses.
Fixes https://github.com/pulumi/pulumi/issues/6004.
The step generator was incorrectly tracking goal states for
old resources, which could lead to a panic if the resource
was removed in the update. This fix only generates goal
states for resources that exist in the updated program.
Now that resources are serialized as refs, ComponentResources
may try to unmarshal local resource refs before they are
initialized during the RegisterResource step.
This change avoids that issue by skipping Output marshaling
for local ComponentResources during the RegisterResource step.
These Outputs will be handled instead during the
RegisterResourceOutputs step.
Co-authored-by: Pat Gavlin <pat@pulumi.com>
The step generator applies `ignoreChanges` pre-processing for all
resources by copying old input values to the new goal for any properties
mentioned in the `ignoreChanges` list. However, this pre-processing
depends on the existence of prior inputs, which by definition does not
exist for a resource being imported prior to the issuance of the
`ImportStep`. These changes add this processing to the implementation of
`ImportStep`, using the inputs read from the provider as the prior
inputs.
- Add component ref coverage to the existing test
- Add coverage for a downlevel SDK communicating with an engine that
supports resource refs
- Add coverage for a downlevel engine communicating with an SDK that
supports resource refs
As part of improving coverage, these changes add a knob to explicitly
disable resource refs in the engine without the use of the environment
variable. The environment variable is now only read by the CLI, and has
been restored to its prior polarity (i.e. `PULUMI_ENABLE_RESOURCE_REFERENCES`).
- Differentiate between resource references that have no ID (i.e. because
the referenced resource is not a CustomResource) and resource references
that have IDs that are not known. This is necessary for proper
backwards-compatible serialization of resource references.
- Fix the key that stores a resource reference's package version in the
.NET, NodeJS, and Python SDKs.
- Ensure that the resource monitor's marshalling/unmarshalling of inputs
and outputs to/from calls to `Construct` retain resource references as
appropriate.
- Fix serialization behavior for resources -> resource references in the
Go SDK: if a resource's ID is unknown, it should still be serialized
as a resource reference, albeit a reference with an unknown ID.
This name better suits the semantics of the type, and aligns with the
rename of deploy.Plan to deploy.Deployment. These changes also refactor
the `update` method s.t. previews and updates are more consistent in
their behavior (e.g. duration and resource changes are now reported for
both, incl. on error paths).
Rename deploy.Plan to deploy.Deployment.
There are two benefits to this change:
1. The name "Deployment" more accurately reflects the behavior of the
type, which is responsible for previewing or executing a deployment.
2. Renaming this type frees up the name "Plan" for use when addressing
#2318.
The way `pulumi new` installs dependencies for .NET projects is slightly different from other languages. For Node.js, Python, and Go, `pulumi new` runs the appropriate command to install project dependencies (e.g. `npm install`, `pip install`, or `go mod download`). For .NET, it calls the same routine used during `preview|up` to ensure required plugins are installed. For .NET, this ends up running `dotnet build` which implicitly installs Nuget packages, builds the project, and also attempts to determine and install the needed Pulumi plugins. When this operation runs during `preview|up`, and there are failures installing a plugin, the error is logged, but deliberately not returned, because an error will be shown for missing plugins later on during the `preview|up` operation. However, during `pulumi new`, we should show any plugin install errors.
Just what it says on the tin. This is implemented by changing the
`GetPackageConfig` method of `ConfigSource` to return a `PropertyMap`
and ensuring that any secret config is represented by a `Secret`.
The langauge SDKs will use this function to fetch the state required to
deserialize a resource reference. SDK support will be added as a follow-up
change.
Contributes to #2430.
Resources are serialized as their URN, ID, and package version. Each
Pulumi package is expected to register itself with the SDK. The package
will be invoked to construct appropriate instances of rehydrated
resources. Packages are distinguished by their name and their version.
This is the foundation of cross-process resources.
Related to #2430.
Co-authored-by: Mikhail Shilkov <github@mikhail.io>
Co-authored-by: Luke Hoban <luke@pulumi.com>
Co-authored-by: Levi Blackstone <levi@pulumi.com>
Move these tests to a new package, `lifecycletest`, that also exposes
APIs that allow consumers to implement their own lifecycle tests. This
is intended to ease the burden of testing plugin implementations and to
set the stage for cleaning up the lifecycle tests themselves.
This involves two changes to the public API, only one of which is
strictly necessary:
- The `host` field of `UpdateOptions` is now exported
- The `Journal` type has been moved from test-only code to the package
proper
The former change is necessary, as it is the mechanism by which package
consumers may inject their own plugin loaders. I was reluctant to expose
this field originally because I wanted to ensure that the behavior of
packages that embed Pulumi is consistent with that of the Pulumi CLI
with respect to plugin loading. I now believe that the risk of consumers
changing this behavior outside of test scenarios is low enough that we
can expose this field. This may also be useful for future scenarios,
e.g. statically linking providers and Pulumi programs.
The latter change is not necessary, but fleshes out the engine package
into a more complete toolkit. Downstream consumers may use the Journal
type to conveniently implement snapshotting.
These changes add support for provider-side previews of create and
update operations, which allows resource providers to supply output
property values for resources that are being created or updated during a
preview.
If a plugin supports provider-side preview, its create/update methods
will be invoked during previews with the `preview` property set to true.
It is the responsibility of the provider to fill in any output
properties that are known before returning. It is a best practice for
providers to only fill in property values that are guaranteed to be
identical if the preview were instead an update (i.e. only those output
properties whose values can be conclusively determined without
actually performing the create/update operation should be populated).
Providers that support previews must accept unknown values in their
create and update methods.
If a plugin does not support provider-side preview, the inputs to a
create or update operation will be propagated to the outputs as they are
today.
Fixes#4992.
* Revise host mode.
The current implementation of host mode uses a `pulumi host` command and
an ad-hoc communication protocol between the engine and client to
connect a language host after the host has begun listening. The most
significant disadvantages of this approach are the communication
protocol (which currently requires the use of stdout), the host-specific
command, and the difficulty of accommodating the typical program-bound
lifetime for an update.
These changes reimplement host mode by adding engine support for
connecting to an existing language runtime service rather than launching
a plugin. This capability is provided via an engine-specific language
runtime, `client`, which accepts the address of the existing languge
runtime service as a runtime option. The CLI exposes this runtime via
the `--client` flag to the `up` and `preview` commands, which similarly
accepts the address of an existing language runtime service as an
argument. These changes also adjust the automation API to consume the
new host mode implementation.
These changes add initial support for the construction of remote
components. For now, this support is limited to the NodeJS SDK;
follow-up changes will implement support for the other SDKs.
Remote components are component resources that are constructed and
managed by plugins rather than by Pulumi programs. In this sense, they
are a bit like cloud resources, and are supported by the same
distribution and plugin loading mechanisms and described by the same
schema system.
The construction of a remote component is initiated by a
`RegisterResourceRequest` with the new `remote` field set to `true`.
When the resource monitor receives such a request, it loads the plugin
that implements the component resource and calls the `Construct`
method added to the resource provider interface as part of these
changes. This method accepts the information necessary to construct the
component and its children: the component's name, type, resource
options, inputs, and input dependencies. It is responsible for
dispatching to the appropriate component factory to create the
component, then returning its URN, resolved output properties, and
output property dependencies. The dependency information is necessary to
support features such as delete-before-replace, which rely on precise
dependency information for custom resources.
These changes also add initial support for more conveniently
implementing resource providers in NodeJS. The interface used to
implement such a provider is similar to the dynamic provider interface
(and may be unified with that interface in the future).
An example of a NodeJS program constructing a remote component resource
also implemented in NodeJS can be found in
`tests/construct_component/nodejs`.
This is the core of #2430.
Certain operations in `engine/diff` mutate engine events during display.
This mutation can occur concurrently with the serialization of the event
for persistence, which causes a panic in the CLI. These changes fix the
offending code and add code that copies each engine event before
persisteing it in order to guard against future issues.
- Remove `Info` from `Source`. This method was not used.
- Remove `Stack` from `EvalSource`. This method was not used.
- Remove `Type` and `URN` from `Step`. These values are available via
`Res().URN.Type()` and `Res().URN`, respectively. This removes the
possibility of inconsistencies between the type, URN, and state of the
resource associated with a `Step`.
- Remove URN from StepEventMetadata.
After importing some resources, and running a second update with the
import still applied, an unexpected replace would occur. This wouldn't
happen for the vast majority of resources, but for some it would.
It turns out that the resources that trigger this are ones that use a
different format of identifier for the import input than they do for the
ID property.
Before this change, we would trigger an import-replacement when an
existing resource's ID property didn't match the import property, which
would be the case for the small set of resources where the input
identifier is different than the ID property.
To avoid this, we now store the `importID` in the statefile, and
compare that to the import property instead of comparing the ID.
