Ensure nothing in the troubleshoot go module depends on consul's top level module. This is so we can import troubleshoot into consul-k8s and not import all of consul.
* turns troubleshoot into a go module [authored by @curtbushko]
* gets the envoy protos into the troubleshoot module [authored by @curtbushko]
* adds a new go module `envoyextensions` which has xdscommon and extensioncommon folders that both the xds package and the troubleshoot package can import
* adds testing and linting for the new go modules
* moves the unit tests in `troubleshoot/validateupstream` that depend on proxycfg/xds into the xds package, with a comment describing why those tests cannot be in the troubleshoot package
* fixes all the imports everywhere as a result of these changes
Co-authored-by: Curt Bushko <cbushko@gmail.com>
The consul container tests orchestrate running containers from various
versions of consul to test things like upgrades. Having the test
framework itself depend on the consul codebase inherently links it to a
specific version of consul which may make some test approaches in the
future difficult.
This change prohibits any such relationship via a custom linting rule.
Unfortunately because the api, sdk, and
test/integration/consul-container packages are submodules of
github.com/hashicorp/consul the gomodguard linter is incapable of
handling those separately hence the need for some custom bash instead.
- remove dep on consul main module
- use 'consul tls' subcommands instead of tlsutil
- use direct json config construction instead of agent/config structs
- merge libcluster and libagent packages together
- more widely use BuildContext
- get the OSS/ENT runner stuff working properly
- reduce some flakiness
- fix some correctness related to http/https API
The fix outlined and merged in #15253 fixed the issue as it occurs in the primary
DC. There is a similar issue that arises when vault is used as the Connect CA in a
secondary datacenter that is fixed by this PR.
Additionally: this PR adds support to run the existing suite of vault related integration
tests against the last 4 versions of vault (1.9, 1.10, 1.11, 1.12)
* add test coverage comments to PRs
* [skip ci] update test coverage
* [skip ci] add .gitattributes to avoid merge conflicts with test coverage
* exempt main and release branches from coverage job
* [skip ci] update test coverage
* [skip ci] update test coverage
* clean up debug line, exit early if missing files
* [skip ci] update test coverage
* extract repository into variable to make porting to ENT easier
* [skip ci] update test coverage
Co-authored-by: hc-github-team-consul-core <github-team-consul-core@hashicorp.com>
Replaces the reflection-based implementation of proxycfg's
ConfigSnapshot.Clone with code generated by deep-copy.
While load testing server-based xDS (for consul-dataplane) we discovered
this method is extremely expensive. The ConfigSnapshot struct, directly
or indirectly, contains a copy of many of the structs in the agent/structs
package, which creates a large graph for copystructure.Copy to traverse
at runtime, on every proxy reconfiguration.
Locally, always run integration tests using amd64, even if running
on an arm mac. This ensures the architecture locally always matches
the CI/CD environment.
In addition:
* Use consul:local for envoy integration and upgrade tests. Previously,
consul:local was used for upgrade tests and consul-dev for integration
tests. I didn't see a reason to use separate images as it's more
confusing.
* By default, disable the requirement that aws credentials are set.
These are only needed for the lambda tests and make it so you
can't run any tests locally, even if you're not running the
lambda tests. Now they'll only run if the LAMBDA_TESTS_ENABLED
env var is set.
* Split out the building of the Docker image for integration
tests into its own target from `dev-docker`. This allows us to always
use an amd64 image without messing up the `dev-docker` target.
* Add support for passing GO_TEST_FLAGs to `test-envoy-integ` target.
* Add a wait_for_leader function because tests were failing locally
without it.
- fix sg: need remote access to test server
- Give the load generator a name
- Update loadtest hcl filename in readme
- Add terraform init
- Disable access to the server machine by default
The upgrade compatibility tests need frequent retrying to pass because they are slightly flaky. To alleviate manual work borrow similar gotestsum logic from the main go tests CI job to rerun them automatically here.
Replace bindata packages with stdlib go:embed.
Modernize some uiserver code with newer interfaces introduced in go 1.16 (mainly working with fs.File instead of http.File.
Remove steps that are no longer used from our build files.
Add Github Action to detect differences in agent/uiserver/dist and verify that the files are correct (by compiling UI assets and comparing contents).
I noticed that the JSON api endpoints for peerings json encodes protobufs directly, rather than converting them into their `api` package equivalents before marshal/unmarshaling them.
I updated this and used `mog` to do the annoying part in the middle.
Other changes:
- the status enum was converted into the friendlier string form of the enum for readability with tools like `curl`
- some of the `api` library functions were slightly modified to match other similar endpoints in UX (cc: @ndhanushkodi )
- peeringRead returns `nil` if not found
- partitions are NOT inferred from the agent's partition (matching 1.11-style logic)
brian shore
Ronald
Valeriia Ruban
Anita Akaeze
Dan Stough
Curt Bushko
Paul Glass
Nitya Dhanushkodi
R.B. Boyer
Kevin Wang
Dhia Ayachi
Chris S. Kim
skpratt
Evan Culver
Dan Upton
Luke Kysow
Chris Thain
John Cowen
cskh
Mark Anderson