This is a genre of invalid output expression that we've seen quite
commonly while testing with 0.11.0-rc1, so we'll call it out specifically
in the upgrade guide and suggest how to fix it.
The differences between the implicit and explicit modes of passing
provider configurations between modules are significant enough to warrant
giving these approaches different names and describing them separately.
This also includes documentation of the current limitation discussed in
#16612, where explicit passing requires a proxy configuration block even
for a _default_ provider configuration, because that limitation is being
accepted for the 0.11.0 release to limit scope.
Now that we're expecting "credentials" blocks in the config (with auth
tokens for private module registries, etc) we should not print out the
config contents into the log, or else people will probably end up
accidentally disclosing their credentials when sharing debug output with
us, or will be reluctant to share debug output.
The provider name coming from ProvidedBy may be resolved if it only
exists in the state. Make sure to strip the module and provider
prefixes for the provider name when adding missing providers.
It's become apparent that passing in a provider config for an implicitly
used provider would be very useful. While the ProviderConfigTransformer
efficiently added providers to the graph, the algorithm was reversed
from what would be needed to allow overriding implicit providers.
Change the ProviderConfigTransformer to fist add all configured
provider, even if they are empty stubs. Then run through all providers
being passed in from the parent, and replace the provider nodes we
created with proxies, and add implicit proxies where none existed. The
extra nodes will then be pruned later.
We are recommending that as of 0.11 all provider configurations be placed
in the root module and, where necessary, be explicitly passed down via
a providers map to customize which configurations are seen by each
child module.
This new section attempts to guide users through such refactoring in the
common case where a child module defines its own provider configuration
based on a value passed in an input variable, and then uses that as
some context to link to the more detailed docs to help those who have
more complex configurations.
The initial pass of this section had some remaining ambiguities, so this
is a second revision that attempts to use terminology more consistently
and to not some additional behaviors that were not described in the
initial version.
We've historically been somewhat inconsistent in how we refer to the
type of object defined by "variable" blocks in configuration. Parts of
our documentation refer to them as "input variables" or just "variables",
while our implementation refers to them as "user variables".
Since Terraform Registry is now also referring to these as "Inputs", here
we standardize on "Input Variable" as the fully-qualified name for this
concept, with "variable" being a shorthand for this where context is
obvious. Outside of this context, anything that can be referred to in
an interpolation expression is generically known as a "variable", with
Input Variables being just one kind, specified by the "var." prefix.
While this terminology shift is not critical yet, it will become more
important as we start to document the new version of the configuration
language so we can use the generic meaning of "variable" there.
The bulk of the text on this page hasn't been revised for some time and
so parts of it were using non-idiomatic terminology or not defining terms
at all.
The main goal of this revision is to standardize on the following terms:
- "provider configuration" refers to a specific provider block in config,
as a distinct idea from the provider _itself_, which is a singleton.
- "Default" vs. "additional" provider configurations, distinguishing
those without and with "alias" arguments respectively. These are named
here so that we can use this terminology to describe the different
behaviors of each for the purposes of provider inheritance between
modules.
There was a bug where all references would be discarded in the case when
a self-reference was encountered. Since a module references all
descendants by it's own path, it returns a self-reference by definition.
Our new resource-to-provider matching is stricter about explicitly
matching aliases when config is present (no longer automatically
inherited) and with locating providers to destroy removed resources.
With this in mind, this is an attempt to expand slightly on this error
message now that users are more likely to see it.
In future it would be nice to do some explicit validation of this a bit
closer to the UI, so we can have room for more explanatory text, but this
additional messaging is intended to help users understand why they might
be seeing this message after removing a provider configuration block from
configuration, whether directly or as a side-effect of removing a module.
Remove the module entry from the state if a module is no longer in the
configuration. Modules are not removed if there are any existing
resources with the module path as a prefix. The only time this should be
the case is if a module was removed in the config, but the apply didn't
target that module.
Create a NodeModuleRemoved and an associated EvalDeleteModule to track
the module in the graph then remove it from the state. The
NodeModuleRemoved dependencies are simply any other node which contains
the module path as a prefix in its path.
This could have probably been done much easier as a step in pruning the
state, but modules are going to have to be promoted to full graph nodes
anyway in order to support count.
You can't find orphans by walking the config, because by definition
orphans aren't in the config.
Leaving the broken test for when empty modules are removed from the
state as well.
Martin Atkins