The Kubernetes provider wanted to return Unimplemented for both
DiffConfig and CheckConfig. However, due to an interaction between the
package we used to construct the error we are returning and the
package we are using to actually construct the gRPC server for the
provider, we ended up in a place where the provider would actually end
up returning an error with code "Unknown", and the /text/ of the
message included information about it being due to the RPC not being
implemented.
So, when we try to call Diff/Check config on the provider, detect this
case as well and treat messages of this shape as if the provider just
returned "Unimplemented".
In 3621c01f4b, we implemented
CheckConfig/DiffConfig incorrectly. We should have explicilty added
the handlers (to supress the warnings we were getting) but returned an
error saying the RPC was not implemented. Instead, we just returned
success but passed back bogus data. This was "fine" at the time
because nothing called these methods.
Now that we are actually calling them, returning incorrect values
leads to errors in grpc. To deal with this we do two things:
1. Adjust the implementations in the dynamic provider to correctly
return not implemented. This allows us to pick up the default engine
behavior going forward.
2. Add some code in CheckConfig/DiffConfig that handle the gRPC error
that is returned when calling methods on the dynamic provider and fall
back to the legacy behavior. This means updating your CLI will not
cause issues for existing resources where the SDK has not been
updated.
For provider plugins, the gRPC interfaces expect that a URN would be
included as part of the DiffConfig/CheckConfig request, which means we
need to flow this value into our Provider interface.
This change does that.
Providers from plugins require that configuration value be
strings. This means if we are passing a secret string to a
provider (for example, trying to configure a kubernetes provider based
on some secret kubeconfig) we need to be careful to remove the
"secretness" before actually making the calls into the provider.
Failure to do this resulted in errors saying that the provider
configuration values had to be strings, and of course, the values
logically where, they were just marked as secret strings
Fixes#2741
We adopt a new algoritm for annotating secrets, which works as
follows:
If the source and destinations are both property maps, annotate their
secrets deeply.
Otherwise, if there is an property in both the input and output arrays
with the same name and the value in the inputs has secrets /anywhere/
in it, mark the output itself a secret.
This means, for example, an array in the inputs with a secret value as
one of the elmenets will mean in the outputs the entire array value is
marked as a secret. This is done because arrays often are treated as
sets by providers and so we really shouldn't consider ordering. It
also means that if a value is added to the array as part of the
operation we still mark the new array as an output even though the
values may not be indentical to the inputs.
For providers which do not natively support secrets (which is all of
them today), we annotate output values coming back from the provider
if there is a coresponding secret input in the inputs we passed in.
This logic was not tearing into rich objects, so if you passed a
secret as a member of an array or object into a resource provider, we
would lose the secretness on the way back.
Because of the interaction with Check (where we call Check and then
take the values returned by the provider as inputs for all calls to
Diff/Update), this would apply not only to the Output values of a
resource but also the Inputs (because the secret metadata would not
flow from the inputs of check to the outputs).
This change augments our logic which transfers secrets metadata from
one property map to another to handle these additional cases.
When a provider does not natively understand secrets, we need to pass
inputs as raw values, as to not confuse it.
This leads to a not great experience by default, where we pass raw
values to `Check` and then use the results as the inputs to remaining
operations. This means that by default, we don't end up retaining
information about secrets in the checkpoint, since the call to `Check`
erases all of our information about secrets.
To provide a nicer experience we were don't lose information about
secrets even in cases where providers don't natively understand them,
we take property maps produced by the provider and mark any values in
them that are not listed as secret as secret if the coresponding input
was a secret.
This ensures that any secret property values in the inputs are
reflected back into the outputs, even for providers that don't
understand secrets natively.
A new `Secret` property value is introduced, and plumbed across the
engine.
- When Unmarshalling properties /from/ RPC calls, we instruct the
marshaller to retain secrets, since we now understand them in the
rest of the engine.
- When Marshalling properties /to/ RPC calls, we use or tracked data
to understand if the other side of the connection can accept
secrets. If they can, we marshall them in a similar manner to assets
where we have a special object with a signiture specific for secrets
and an underlying value (which is the /plaintext/ value). In cases
where the other end of the connection does not understand secrets,
we just drop the metadata and marshal the underlying value as we
normally would.
- Any secrets that are passed across the engine events boundary are
presently passed as just `[secret]`.
- When persisting secret values as part of a deployment, we use a rich
object so that we can track the value is a secret, but right now the
underlying value is not actually encrypted.
In previous commits, we have changed the language plugin protocol to
allow the host to communicate that the plugin is meant to boot in "query
mode." In nodejs, this involves not doing things like registering the
default stack resource. This commit will implement this functionality.
This commit re-uses an error reporting mechanism previously used when
the plugin loader fails to locate a plugin that is compatible with the
requested plugin version. In addition to specifying what version we
attempted to load, it also outputs a command that will install the
missing plugin.
These changes take advantage of the newly-added support for returning
inputs from Read to update a resource's inputs as part of a refresh.
As a consequence, the Pulumi engine will now properly detect drift
between the actual state of a resource and the desired state described
by the program and generate appropriate update or replace steps.
As part of these changes, a resource's old inputs are now passed to the
provider when performing a refresh. The provider can take advantage of
this to maintain the accuracy of any additional data or metadata in the
resource's inputs that may need to be updated during the refresh.
This is required for the complete implementation of
https://github.com/pulumi/pulumi-terraform/pull/349. Without access to
the old inputs for a resource, TF-based providers would lose all
information about default population during a refresh.
If a provider returns information about the top-level properties that
differ, use those keys to filter the diffs that are rendered to the
user.
Fixes#2453.
We previously logged the number of replaces and changes returned from a call to Diff, but not the actual properties that were forcing replace. Several times we've had to debug issues with unexpected replaces being proposed, and this information is very useful to have access to.
Changes the verbose logging to include the property names for both replaces and changes instead of just the count.
- Add support for per-property dependencies to the Go SDK
- Add tests for first-class secret rejection in the checkpoint and RPC
layers and language SDKs
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.
After #2088, we began calling `Diff` on providers that are not configured
due to unknown configuration values. This hit an assertion intended to
detect exactly this scenario, which was previously unexpected.
These changes adjust `Diff` to indicate that a Diff is unavailable and
return an error message that describes why. The step generator then
interprets the diff as indicating a normal update and issues the error
message to the diagnostic stream.
Fixes#2223.
We run the same suite of changes that we did on gometalinter. This
ended up catching a few new issues, some of which were addressed and
some of which were baselined.
In preparation for some workspace restructuring, I decided to scratch a
few itches of my own in the code:
* Change project's RuntimeInfo field to just Runtime, to match the
serialized name in JSON/YAML.
* Eliminate the no-longer-used Context and NoDefaultIgnores fields on
project, and all of the associated legacy PPC-related code.
* Eliminate the no-longer-used IgnoreFile constant.
* Remove a bunch of "// nolint: lll" annotations, and simply format
the structures with comments on dedicated lines, to avoid overly
lengthy lines and lint suppressions.
* Mark Dependencies and InitErrors as `omitempty` in the JSON
serialization directives for CheckpointV2 files. This was done for
the YAML directives, but (presumably accidentally) omitted for JSON.
Go 1.10 made some breaking changes to the headers in archive/tar [1] and
archive/zip [2], breaking the expected values in tests. In order to keep
tests passing with both, wherever a hardcoded hash is expect we switch
on `runtime.Version()` to select whether we want the Go 1.9 (currently
supported Go version) or later version of the hash.
Eventually these switches should be removed in favour of using the later
version only, so they are liberally commented to explain the reasoning.
[1]: https://golang.org/doc/go1.10#archive/tar
[2]: https://golang.org/doc/go1.10#archive/zip
Providers with unknown properties are currently considered to require
replacement. This was intended to indicate that we could not be sure
whether or not replacement was reqiuired. Unfortunately, this was not a
good user experience, as replacement would never be required at runtime.
This caused quite a bit of confusion--never proposing replacement seems
to be the better option.
* Protobuf changes
* Move management of root resource state to engine
This commit fixes a persistent side-by-side issue in the NodeJS SDK by
moving the management of root resource state to the engine. Doing so
adds two new endpoints to the Engine gRPC service: 1) GetRootResource
and 2) SetRootResource, which get and set the root resource
respectively.
* Rebase against master, regenerate proto
We signal provider cancellation by hangning a goroutine off of the plan
executor's parent context. To ensure clean shutdown, this goroutine also
listens on a channel that closes once the plan has finished executing.
Unfortunately, we were closing this channel too early, and the close was
racing with the cancellation signal. These changes ensure that the
channel closes after the plan has fully completed.
Fixes#1906.
Fixespulumi/pulumi-kubernetes#185.
This commit reverts most of #1853 and replaces it with functionally
identical logic, using the notion of status message-specific sinks.
In other words, where the original commit implemented ephemeral status
messages by adding an `isStatus` parameter to most of the logging
methdos in pulumi/pulumi, this implements ephemeral status messages as a
parallel logging sink, which emits _only_ ephemeral status messages.
The original commit message in that PR was:
> Allow log events to be marked "status" events
>
> This commit will introduce a field, IsStatus to LogRequest. A "status"
> logging event will be displayed in the Info column of the main
> display, but will not be printed out at the end, when resource
> operations complete.
>
> For example, for complex resource initialization, we'd like to display
> a series of intermediate results: [1/4] Service object created, for
> example. We'd like these to appear in the Info column, but not at the
> end, where they are not helpful to the user.
The glog package force the use of golang's underyling flag package,
which Cobra does not use. To work around this, we had a complicated
dance around defining flags in multiple places, calling flag.Parse
explicitly and then stomping values in the flag package with values we
got from Cobra.
Because we ended up parsing parts of the command line twice, each with
a different set of semantics, we ended up with bad UX in some
cases. For example:
`$ pulumi -v=10 --logflow update`
Would fail with an error message that looked nothing like normal CLI
errors, where as:
`$ pulumi -v=10 update --logflow`
Would behave as you expect. To address this, we now do two things:
- We never call flag.Parse() anymore. Wacking the flags with values we
got from Cobra is sufficent for what we care about.
- We use a forked copy of glog which does not complain when
flag.Parse() is not called before logging.
Fixes#301Fixes#710Fixes#968
1. 'readID' was never assigned to and was always the default value,
leading the refresh source to believe a resource was deleted
2. The refresh source could hang when a resource is deleted.
Some time ago, we introduced the concept of the initialization error to
Pulumi (i.e., an error where the resource was successfully created but
failed to fully initialize). This was originally implemented in `Create`
and `Update` methods of the resource provider interface; when we
detected an initialization failure, we'd pack the live version of the
object into the error, and return that to the engine.
Omitted from this initial implementation was a similar semantics for
`Read`. There are many implications of this, but one of them is that a
`pulumi refresh` will erase any initialization errors that had
previously been observed, even if the initialization errors still exist
in the resource.
This commit will introduce the initialization error semantics to `Read`,
fixing this issue.
### 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`.
This change lets us set runtime specific options in Pulumi.yaml, which
will flow as arguments to the language hosts. We then teach the nodejs
host that when the `typescript` is set to `true` that it should load
ts-node before calling into user code. This allows using typescript
natively without an explicit compile step outside of Pulumi.
This works even when a tsconfig.json file is not present in the
application and should provide a nicer inner loop for folks writing
typescript (I'm pretty sure everyone has run into the "but I fixed
that bug! Why isn't it getting picked up? Oh, I forgot to run tsc"
problem.
Fixes#958
When a resource fails to initialize (i.e., it is successfully created,
but fails to transition to a fully-initialized state), and a user
subsequently runs `pulumi update` without changing that resource, our
CLI will fail to warn the user that this resource is not initialized.
This commit begins the process of allowing our CLI to report this by
storing a list of initialization errors in the checkpoint.
This commit adds CLI support for resource providers to provide partial
state upon failure. For resource providers that model resource
operations across multiple API calls, the Provider RPC interface can now
accomodate saving bags of state for resource operations that failed.
This is a common pattern for Terraform-backed providers that try to do
post-creation steps on resource as part of Create or Update resource
operations.
