There are a few constructs from 0.11 and prior that cause 0.12 parsing to
fail altogether, which previously created a chicken/egg problem because
we need to install the providers in order to run "terraform 0.12upgrade"
and thus fix the problem.
This changes "terraform init" to use the new "early configuration" loader
for module and provider installation. This is built on the more permissive
parser in the terraform-config-inspect package, and so it allows us to
read out the top-level blocks from the configuration while accepting
legacy HCL syntax.
In the long run this will let us do version compatibility detection before
attempting a "real" config load, giving us better error messages for any
future syntax additions, but in the short term the key thing is that it
allows us to install the dependencies even if the configuration isn't
fully valid.
Because backend init still requires full configuration, this introduces a
new mode of terraform init where it detects heuristically if it seems like
we need to do a configuration upgrade and does a partial init if so,
before finally directing the user to run "terraform 0.12upgrade" before
running any other commands.
The heuristic here is based on two assumptions:
- If the "early" loader finds no errors but the normal loader does, the
configuration is likely to be valid for Terraform 0.11 but not 0.12.
- If there's already a version constraint in the configuration that
excludes Terraform versions prior to v0.12 then the configuration is
probably _already_ upgraded and so it's just a normal syntax error,
even if the early loader didn't detect it.
Once the upgrade process is removed in 0.13.0 (users will be required to
go stepwise 0.11 -> 0.12 -> 0.13 to upgrade after that), some of this can
be simplified to remove that special mode, but the idea of doing the
dependency version checks against the liberal parser will remain valuable
to increase our chances of reporting version-based incompatibilities
rather than syntax errors as we add new features in future.
* command/show: added test scaffold for json output
More test cases will be added once the basic shape of the tests is
validated.
- command/json* packages now sort resources by address, matching
behavior elsewhere
- using cmp in tests instead of reflect.DeepEqual for the diffs
- updating expected output in tests to match sorting
Previously we were doing this rather inconsistently: some commands would
do it and others would not. By doing it here we ensure we always apply the
same normalization, regardless of which operation we're running.
This normalization is mostly for cosmetic purposes in error messages, but
it also ends up being used to populate path.module and path.root and so
it's important that we always produce consistent results here so that
we don't produce flappy changes as users work with different commands.
The fact that thus mutates a data structure as a side-effect is not ideal
but this is the best place to ensure it always gets applied without doing
any significant refactoring, since everything after this point happens in
the backend package where the normalizePath method is not available.
* command/show: adding functions to aid refactoring
The planfile -> statefile -> state logic path was getting hard to follow
with blurry human eyes. The getPlan... and getState... functions were
added to help streamline the logic flow. Continued refactoring may follow.
* command/show: use ctx.Config() instead of a config snapshot
As originally written, the jsonconfig marshaller was getting an error
when loading configs that included one or more modules. It's not clear
if that was an error in the function call or in the configloader itself,
but as a simpler solution existed I did not dig too far.
* command/jsonplan: implement jsonplan.Marshal
Split the `config` portion into a discrete package to aid in naming
sanity (so we could have for example jsonconfig.Resource instead of
jsonplan.ConfigResource) and to enable marshaling the config on it's
own.
Older versions of terraform could save the backend hash number in a
value larger than an int.
While we could conditionally decode the state into an intermediary data
structure for upgrade, or detect the specific decode error and modify
the json, it seems simpler to just decode into the most flexible value
for now, which is a uint64.
A lot of commands used `c.Meta.flagSet()` to create the initial flagset for the command, while quite a few of them didn’t actually use or support the flags that are then added.
So I updated a few commands to use `flag.NewFlagSet()` instead to only add the flags that are actually needed/supported.
Additionally this prevents a few commands from using locking while they actually don’t need locking (as locking is enabled as a default in `c.Meta.flagSet()`.
Next to adding the locking for the `state push` command, this commit also fixes a small bug where the lock would not be propertly released when running the `state show` command.
And finally it renames some variables in the `[un]taint` code in order to try to standardize the var names of a few frequently used variables (e.g. statemgr.Full, states.State, states.SyncState).
In a couple places in tests we execute a child "go build" to make a helper
program. Now that we're running in module mode, "go build" will normally
default to downloading and caching dependencies, which we don't want
because we're still using vendoring for the moment.
Therefore we need to instruct these child builds to use vendoring too,
avoiding the need to download all of the dependencies and ensuring that
we'll be building with the same dependencies that we'd use for a normal
build.
Several of these tests rely on external services (e.g. Terraform Registry)
that have not yet been updated to support the needs of Terraform v0.12.0,
so for now we'll skip all of these tests and wait until those systems have
been updated.
This should be removed before Terraform v0.12.0 final to enable these
tests to be used as part of pre-release smoke testing.
The local filesystem state manager no longer creates backup files eagerly,
instead creating them only if on first write there is already a snapshot
present in the target file.
Therefore for this test to exercise the codepaths it intends to we must
create an initial state snapshot for it to overwrite, creating the backup
in the process.
There are several other tests for this behavior elsewhere, so this test
is primarily to verify that the refresh command is configuring the backend
appropriately to get the backups written in the desired location.
We now only create a backup state file if the given output file already
exists, which it does not in this test.
(The behavior of creating the backup files is already covered by other
tests, so no need for this one go out of its way to do it.)
We now don't create a local state backup until the first snapshot write,
so we don't expect there to be a backup file until the end of the test.
(There is already a check at the end there, unmodified by this change.)
The filesystem backend has the option of using a different file for its
initial read.
Previously we were incorrectly writing the contents of that file out into
the backup file, rather than the prior contents of the output file. Now
we will always read the output file in RefreshState in order to decide
what we will back up but then we will optionally additionally read the
input file and prefer its content as the "current" state snapshot.
This is verified by command.TestMetaBackend_planLocalStatePath and
TestMetaBackend_configureNew, which are both now passing.
The changes to how we handle setting the state path on the local backend
broke the heuristic we were using here for detecting migration from one
local backend to another with the same state path, which would by default
end up deleting the state altogether after migration.
We now use the StatePaths method to do this, which takes into account
both the default values and any settings that have been set.
Additionally this addresses a flaw in the old method which could
potentially have deleted all non-default workspace state files if the
"path" setting were changed without also changing the "workspace_dir"
setting. This new approach is conservative because it will preserve all
of the files if any one overlaps.
In an earlier change we fixed the "backendFromConfig" codepath to be
able to properly detect changes to the -backend-config arguments during
"terraform init", but this detection is too strict for the normal case
of running an operation in a previously-initialized directory.
Before any of the recent changes, the logic here was to selectively update
the hash to include -backend-config settings in the init case. Since
that late hash recalculation was confusing, here we take the alternative
path of using the hash only in the normal case and full value comparison
in the init case. Treating both of these cases separately makes things
marginally easier to follow here.
The import command was imposing the default state path at the CLI level,
rather than leaving that to be handled by the backend. As a result, the
output state was always forced to be terraform.tfstate, regardless of
the backend settings.
This test is testing some strange implementation details of the old
local backend which do not hold with the new filesystem state manager.
Specifically, it was expecting state to be read from the stateOutPath
rather than the statePath, which makes no sense here because the backend
is configured to read from the default terraform.tfstate file (which does
not exist.)
There is another problem with this test which will be addressed in a
subsequent commit.
As part of integrating the new "remote" backend we relaxed the requirement
that a "default" workspace must exist in all backends and now skip
migrating empty workspace states to avoid creating unnecessary "default"
workspaces when switching between backends that require it and backends
that don't, such as when switching from the local backend (which always
has a "default" workspace) to Terraform Enterprise.
This was failing because we now handle the settings for the local backend
a little differently as a result of decoding it with the HCL2 machinery.
Specifically, the backend.State* fields are now assumed to be what is
given in configuration, and any CLI overrides are maintained separately
in OverrideState* fields so that they can be imposed "just in time" in
StatePaths.
This is particularly important because OverrideStatePath (when set) is
used regardless of workspace name, while StatePath is a suitable value
only for the "default" workspace, with others needing to be constructed
from StateWorkspaceDir instead.
Our new state model has a different implementation of "empty" that doesn't
consider lineage/serial, so we need to have some actual content in these
state fixtures to avoid them being skipped during state migrations.
We previously hacked around the import/export functionality being missing
in the statemgr layer after refactoring, but now it's been reintroduced
to fix functionality elsewhere we should use the centralized Import and
Export functions to ensure consistent behavior.
In particular, this pushes the logic for checking lineage and serial
during push down into the state manager itself, which is better because
all other details about lineage and serial are managed within the state
managers.
This test was initially failing because its fixture had a state which our
new state models consider to be "empty", and thus it was not migrated.
After fixing that (by adding an output to the fixture), this revealed a
bug that the lineage was not being persisted through the migration. This
is fixed by using the statemgr.Migrate method instead of writing via the
normal Writer interface, which allows two cooperating state managers to
properly transfer the lineage and serial along with the state snapshot.
This test was incorrectly updated in a previous iteration, with it
creating a modified state to write but then not actually writing it,
writing an empty test state instead.
This made the test fail because a backup state file is created only if
the new state snapshot is different to the old when written.
Some other test is leaving behind a terraform.tfstate after it concludes,
which can cause this test to fail in a strange way due to picking up
extra provider requirements from that state.
This check doesn't fix that problem, but it at least makes the test fail
in a more helpful way to avoid time wasted trying to debug this test when
it's some other test that actually has the bug.
This test is currently failing due to the command completing successfully,
which would previously cause a panic because we didn't properly initialize
the MockUi and so its error buffer is nil unless written to.
(The failure this was masking will be fixed in a subsequent commit.)
In prior refactoring we lost the required core version check from
"terraform init", which we restore here.
Additionally, this test used to have an incorrect name that suggested it
was testing something in the "getProvider" codepath, but version checking
happens regardless of what other options are selected.
After all of the refactoring we were no longer checking the Terraform
version field in a state file, causing this test to fail.
This restores that check, though with a slightly different error message.
This test was using old-style state files as its input, differing only by
lineage. Since lineages are now managed within the state manager itself,
the test can't use that to distinguish the two files and so we put a
different output in each one instead.
This also introduces some TRACE logging to the migration codepaths.
There's some hard-to-follow control flow here and so this extra logging
helps to understand the reason for a particular outcome, and since this
codepath is visited only in "terraform init" anyway it doesn't hurt to
be a bit more verbose here.
In the refactoring for new HCL this codepath stopped taking into account
changes to the CLI -backend-config options when deciding if a backend
migration is required.
This restores that behavior in a different way than it used to be: rather
than re-hashing the merged config and comparing the hashes, we instead
just compare directly the configuration values, which must be exactly
equal in order to skip migration.
This change is covered by the test TestInit_inputFalse, although as of
this commit it is still not passing due a downstream problem within the
migration code itself.
This test was re-using the same InitCommand value to run multiple times,
which is not realistic. Since we now cache configuration source code
inside command.Meta on load, it's important that we use a fresh
InitCommand instance here so it'll see the modified configuration file
we've left on disk.
Here we were going to the trouble of copying the body so we could mutate
it, but then ended up mutating the original anyway and then returning the
unmodified copy. Whoops!
This fix is verified by a number of "init" command tests that exercise the
-backend-config option, including TestInit_backendConfigFile and several
others whose names have the prefix TestInit_backendConfig .
When we originally wrote this message we struggled a bit for how to refer
to the releases server without writing an awkwardly-ungrammatical
sentence, and so "the official repository" became a placeholder name for
it.
Now that we'll be looking in Terraform Registry this gives us a nice
proper noun to use. This message will need to evolve more as our
integration with the registry gets more sophisticated, but for now this
works.
Some over-zealous bulk updating of this test file caused this test to be
producing a remote state config cache file on disk when it doesn't
actually need one: the backend config comes from the plan file when
applying a saved plan.
Some merging conflict shenanigans here led to this usage not lining up
with the imported symbol name, meaning that the tests couldn't compile any
more.
We missed fixing this up during the big updates for the new plan/state
models since the failures were being masked by testBackendState being
broken.
This is the same sort of update made to many other tests: add schema to
the mock provider, adjust for the new plan/state types, and make
allowances for the new built-in diffing behavior in core.
The hashing function for cached backend configuration is different now, so
our hard-coded hash of the HTTP backend address wasn't working anymore.
Here we update the hash so that tests using this test backend will work
again. Rather than leaving it hard-coded, we'll instead compute it the
same way as "terraform init" would.
In practice only one test is actually using this function right now, so
we also update the test fixture for that test (TestPlan_outBackend) to
match the new expectations, though as of this commit it's still failing
with an unrelated error.
The mission of this process method used to include dealing with
auto-loaded tfvars files, but it doesn't do that anymore.
It does still deal with the -no-color option, but the test wasn't
exercising that part before.
Now the test here focuses on the -no-color behavior.
The process method still has a "vars" flag argument which is no longer
used. Since this is an unexported method we could potentially address this
but this commit is intentionally limited only to fixing the test.
Comments here indicate that this was erroneously returning an error but
we accepted it anyway to get the tests passing again after other work.
The tests over in the "terraform" package agree that cancelling should be
a successful outcome rather than an error.
I think that cancelling _should_ actually be an error, since Terraform did
not complete the operation it set out to complete, but that's a change
we'd need to make cautiously since automation wrapper scripts may be
depending on the success-on-cancel behavior.
Therefore this just fixes the command package test to agree with the
Terraform package tests and adds some FIXME notes to capture the potential
that we might want to update this later.
The State.Equal function is now more precise than this test needs. It's
only trying to distinguish between an empty state and a non-empty state,
so the string representation of state is good enough to get that done
while disregarding other subtle differences.
This new source type should be used for variables loaded from .tfvars files that were explicitly passed as command line arguments (e.g. -var-file=foo.tfvars)
Without using absolute paths any module info is lost in the output. And the attributes were randomly ordered and so changed between different executions of the command.
When HCL encounters an error during expression evaluation, it annotates
its diagnostics with information about the expression that was being
evaluated and the EvalContext it was evaluated in.
This gives us enough information to show helpful hints to the user about
the final values of any reference expressions that are present in the
expression, which is very useful extra context for expressions that get
evaluated multiple times, such as:
- Any expression in a block with "count" or "for_each" set
- The sub-expressions within a "for" expression
This work was done against APIs that were already changed in the branch
before work began, and so it doesn't apply to the v0.12 development work.
To allow v0.12 to merge down to master, we'll revert this work out for now
and then re-introduce equivalent functionality in later commits that works
against the new APIs.
This reinstates an old behavior that was lost in the reorganization of how
we deal with the -var and -var-file options.
This fix is verified by TestApply_planVars now passing.
In the new implementation of collecting variables I initially forgot the
JSON variant of terraform.tfvars.
This fix is verified by TestApply_varFileDefaultJSON now passing.
This was previously targeting the old state manager and state types, so it
needed some considerable rework to get it working again with the new state
types.
Since our new resource address syntax lacks the weird extra .deposed
special case we had before, we instead interpret addresses as
whole-instance addresses here and remove the deposed objects along with
the current one (if present), since this is more likely to match with
user expectations because we don't consider deposed objects to be
independently addressable in any other situation.
With that said, to be more explicit about what is going on we do now have
a -dry-run mode and maintain separate counts of current and deposed
instances so that we can expose that in the UI where relevant.
Radek Simko