connector acceptance tests: bug fixes #44529
Conversation
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎ 1 Skipped Deployment
|
| cursor_field=[], | ||
| minimum_generation_id=1, | ||
| generation_id=1, | ||
| sync_id=1, |
There was a problem hiding this comment.
Without this fix, source connectors built with the bulk CDK will fail unexpectedly. The Bulk CDK performs schema validation on the connector CLI inputs, so if these fields are missing, its execution fails very early.
There was a problem hiding this comment.
Would you mind adding a comment with a link to the bulk CDK validation?
| @@ -54,7 +55,7 @@ async def _run_with_config(container: dagger.Container, command: List[str], conf | |||
|
|
|||
|
|
|||
| async def _run(container: dagger.Container, command: List[str]) -> dagger.Container: | |||
| return await container.with_exec(command, skip_entrypoint=True) | |||
| return await (container.with_env_variable("CACHEBUSTER", str(uuid.uuid4())).with_exec(command, skip_entrypoint=True)) | |||
There was a problem hiding this comment.
If we don't add this cache-busting layer, the with_exec layer itself can be cached. This became apparent when the container setup auto-fixture didn't actually run like I thought it would (because the run was identical to another, earlier test case)
| cursor_field=[], | ||
| minimum_generation_id=1, | ||
| generation_id=1, | ||
| sync_id=1, |
There was a problem hiding this comment.
Would you mind adding a comment with a link to the bulk CDK validation?
What
Bug fixes for dagger execution caching and for failure trace message test case in the connector acceptance test suite.
User Impact
Decreased chance of false positive CAT test runs.
Can this PR be safely reverted and rolled back?