Commit Graph

1154 Commits

Author SHA1 Message Date
findkim 7816e61614 Bump installer protocol version to 5 and separate client and server protocol references 2019-01-16 15:07:57 -06:00
Radek Simko 13896d72c5
command/format: Render empty object as {} 2019-01-15 14:34:49 +00:00
Radek Simko 5f14b7a7f2
command/format: Render empty primitive list/set as [] 2019-01-15 14:34:49 +00:00
Radek Simko 09d19ca9d9
command/format: Render empty JSON object as {} 2019-01-15 14:34:48 +00:00
Radek Simko 0dd2d56f18
command/format: Render empty maps as {} 2019-01-15 14:34:46 +00:00
Kim Ngo 41a2376915
Merge pull request #19981 from findkim/colorize-protocol-mismatch-msg
Condense protocol mismatch UI error message
2019-01-14 15:12:09 -06:00
findkim 6e0de3e3f5 Nest incompatible provider protocol error to include dynamic custom msg 2019-01-14 14:56:47 -06:00
Radek Simko aae87c7921
Merge pull request #19990 from hashicorp/f-cmd-fmt-ignore-empty-str
command/format: Ignore removal of empty strings
2019-01-14 20:33:43 +00:00
Martin Atkins 86c02d5c35 command: "terraform init" can partially initialize for 0.12upgrade
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.
2019-01-14 11:33:21 -08:00
Martin Atkins 0c0a437bcb Move module install functionality over to internal/initwd 2019-01-14 11:33:21 -08:00
Radek Simko d96f4fa77b
command/format: Ignore removal of empty strings 2019-01-13 22:56:04 +00:00
Kristin Laemmert 5df9cd0f52
command/show: tests for -json output (#19980)
* 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
2019-01-11 15:13:55 -08:00
Radek Simko bc4b7cad68
command/format: Render null in dark gray (#19616) 2019-01-11 19:27:09 +00:00
Kristin Laemmert cdf7cc2449
command/json*: updating documentation and adding tests (#19944)
A few minor fixes and cleanups as a result of said tests. Hooray for
eventual consistency!
2019-01-09 08:59:11 -08:00
Sander van Harmelen fb57e9b26f command/state: add proper locking 2019-01-08 14:57:52 +01:00
Kristin Laemmert 3e9c51c726
command/jsonstate: marshal state into a machine readable blob o'json (#19911)
* command/jsonstate: marshal state into a machine readable blob o'json
* consistent json tags are nice
2019-01-03 12:08:03 -08:00
Kristin Laemmert 2b6dc13f29
jsonplan: remove "proposed_unknown" structure in favor of "after_unknown" field in "change" (#19709)
* jsonplan: remove "proposed_unknown" structure in favor of
"after_unknown" field in "change"
2018-12-20 14:30:18 -08:00
Martin Atkins 2cf63d068f command: Always normalize config path before operations
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.
2018-12-19 13:47:42 -08:00
Kristin Laemmert 126e5f337f
json output of terraform plan (#19687)
* 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.
2018-12-19 11:08:25 -08:00
James Bardin 149ccd929e missing commits from 19688 2018-12-18 16:54:09 -08:00
James Bardin 9667e06a03 decode backend hash as uint64
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.
2018-12-18 17:57:44 -05:00
Charles Kenney 2dfc3a399d command: allow -no-color option on "providers" command 2018-12-17 10:55:16 -08:00
Radek Simko f64978b64c
backend/local: Render CBD replacement (+/-) correctly (#19642)
* backend/local: Render CBD replacement (+/-) correctly

* command/format: Use IsReplace helper function
2018-12-14 13:45:47 +00:00
Radek Simko b28efa0bf3
command/format: Fix tests 2018-12-12 14:28:12 +00:00
Radek Simko 0b981fa641
command/format: Fix rendering of force-new updates 2018-12-12 14:19:17 +00:00
Radek Simko fe117e9f02
command/format: Fix rendering of nested blocks during update 2018-12-11 18:07:21 +00:00
Radek Simko 2df886397b
command/format: Add more tests to cover non-primitive fields 2018-12-11 14:16:25 +00:00
Radek Simko e3e459a8d4
command/format: Restructure tests 2018-12-11 14:16:17 +00:00
Radek Simko 49e7026bdd
command/format: Add tests for ResourceChange renderer 2018-12-10 17:42:45 +00:00
Kristin Laemmert e3cecb7d56
format/state: added missing newline in the `outputs` output (#19542) 2018-12-04 11:37:22 -08:00
Radek Simko 9e5677adeb
Merge pull request #19453 from hashicorp/b-whole-body-diag-rendering
command/format: Fix rendering of attribute-agnostic diagnostics
2018-11-26 23:52:21 +00:00
Radek Simko 501fcd1e21
Fix tests after upgrading hcl 2018-11-26 23:38:37 +00:00
Radek Simko 4f2868a019
command/format: Fix rendering of attribute-agnostic diagnostics 2018-11-26 23:38:37 +00:00
Sander van Harmelen ef9054562e commands: make sure the correct flagset is used
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()`.
2018-11-23 16:13:34 +01:00
Sander van Harmelen 79a9a15879 command/state: lock when pushing state
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).
2018-11-20 11:15:16 +01:00
Martin Atkins 884aa387b8 command: Use vendoring when building helper programs in tests
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.
2018-11-19 11:41:52 -08:00
Martin Atkins 5255e85238 "go fmt" fixups
Apparently my editor is still not reliably formatting on save, so I missed
a few formatting quirks in these files.
2018-11-19 09:02:35 -08:00
Martin Atkins 73c9521a04 command/e2etest: Temporarily disable tests that access network
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.
2018-11-19 09:02:35 -08:00
Martin Atkins 37bc187f95 command: "terraform output" mustn't panic when no state is present
This is verified by TestOutput_noArgs.
2018-11-19 09:02:35 -08:00
Martin Atkins 53b5b95ef5 command: Fix TestRefresh_backup
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.
2018-11-19 09:02:35 -08:00
Martin Atkins 762a173c7f command: Fix TestRefresh_outPath
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.)
2018-11-19 09:02:35 -08:00
Martin Atkins e2ba90fdfa command: Fix TestMetaBackend_planLocalMatch
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.)
2018-11-19 09:02:35 -08:00
Martin Atkins 48601d261d states/statemgr: In Filesystem, back up output file, not input file
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.
2018-11-19 09:02:35 -08:00
Martin Atkins 27abd9c6b8 command: Fix TestMetaBackend_localDoesNotDeleteLocal
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.
2018-11-19 09:02:35 -08:00
Martin Atkins be79bf0412 command: Fix TestPlan_outBackend
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.
2018-11-19 09:02:35 -08:00
Martin Atkins b316e4ab56 command: Fix TestImport_remoteState
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.
2018-11-19 09:02:35 -08:00
Martin Atkins 2b9f92be31 command: Partially fix TestMetaBackend_planLocalStatePath
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.
2018-11-19 09:02:35 -08:00
Martin Atkins aecb66d3db command: Fix TestMetaBackend_configuredChangeCopy_multiToNoDefaultWithoutDefault
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.
2018-11-19 09:02:35 -08:00
Martin Atkins ec27526cc3 command: Fix TestMetaBackend_configuredChangeCopy_multiToMulti
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.
2018-11-19 09:02:35 -08:00
Martin Atkins 6c7cecfbd8 command: More logging during migration
This just finishes off the logging added in earlier commits to get all
the way through to the actual migration call.
2018-11-19 09:02:35 -08:00