Populate __ENV.K6_CLOUDRUN_TEST_RUN_ID on local executions of k6 cloud run #4092
Conversation
|
Note that the approach followed here won't make But I'd be very happy to see any suggestions or ideas for possible workarounds. |
cmd/outputs_cloud.go
Outdated
| raw, err := cloudConfToRawMessage(conf) | ||
| if err != nil { | ||
| return fmt.Errorf("could not serialize cloud configuration: %w", err) | ||
| } | ||
|
|
||
| test.derivedConfig.Collectors[builtinOutputCloud.String()] = raw | ||
|
|
||
| return nil |
There was a problem hiding this comment.
Is this actually needed ? I guess maybe if we save it to disk, but not otherwise?
There was a problem hiding this comment.
I'm lacking historical context here (about when and how those overrides are used), but in case there's overrides from Cloud, I guess we might want those to be applied and "sent" to the Cloud output 🤔
If not, I'm happy with removing the entire if block and just ignore overrides from Cloud here.
|
@joanlopez maybe first rebase though as there have been a few changes including to the CI that mean theat you will have more lint failures. |
joanlopez
changed the title
|
Please @andrewslotin, @mstoykov, That's mainly because that would require a larger refactor, to completely move the creation of the test run outside of the Cloud output, as I did in master...k6/cloud, but that would mean "making the Cloud output even a more special output", and @mstoykov discouraged that. Otherwise, I don't think there's any way we can modify the environment from within the Cloud output particularly, and reflect that to the test's environment. But I'm open for suggestions/ideas, if any. PS: @olegbespalov adding you as a reviewer, cause I think @oleiade won't be able to follow up on this PR for now. |
joanlopez
force-pushed
the
k6/cloud-v3
branch
from
f891588 to
893b760
Compare
|
@joanlopez, since |
I agree, I just wanted to make it clear so w'all set proper expectations. |
cmd/cloud_run.go
Outdated
| // If the "K6_CLOUDRUN_TEST_RUN_ID" is set, then it means that this code is being executed in the k6 Cloud. | ||
| // Therefore, we don't need to continue with the test run creation, as we don't need to create any test run. | ||
| // This should technically never happen, as k6, when executed in the Cloud, it uses the standard "run" | ||
| // command "locally", but we add this early return just in case, for safety. | ||
| // | ||
| // If not, we know this execution requires a test run to be created in the Cloud. | ||
| // So, we create it as part of the process of loading and configuring the test. | ||
| if _, isSet := c.runCmd.gs.Env[testRunIDKey]; !isSet { |
There was a problem hiding this comment.
I think we should just drop this part for now and maybe add it later.
cmd/cloud_run.go
Outdated
| // If the "K6_CLOUDRUN_TEST_RUN_ID" is set, then it means that this code is being executed in the k6 Cloud. | ||
| // Therefore, we don't need to continue with the test run creation, as we don't need to create any test run. | ||
| // This should technically never happen, as k6, when executed in the Cloud, it uses the standard "run" | ||
| // command "locally", but we add this early return just in case, for safety. | ||
| // | ||
| // If not, we know this execution requires a test run to be created in the Cloud. | ||
| // So, we create it as part of the process of loading and configuring the test. | ||
| if _, isSet := c.runCmd.gs.Env[testRunIDKey]; !isSet { |
Co-authored-by: Oleg Bespalov <[email protected]>
What?
It populates the
__ENV.K6_CLOUDRUN_TEST_RUN_ID, which is the environment variable that contains the test run id on Cloud executions, with the corresponding value for local executions streaming results to the Cloud:k6 cloud run --local-execution.Why?
Because this unlocks different projects we have ongoing, currently blocked by the lack of a mechanism to read that variable on local executions streaming results to the Cloud.
Checklist
make lint) and all checks pass.make tests) and all tests pass.Related PR(s)/Issue(s)
Contributes to #4041