test(spanner): fix flaky restore_database() sample test #4641
Conversation
google-cla
bot
added
the
cla: yes
larkee
added
the
kokoro:force-run
kokoro-team
removed
the
kokoro:force-run
product-auto-label
bot
added
the
samples
larkee
changed the title
| @@ -67,6 +67,13 @@ def test_create_backup(capsys, database): | |||
| assert BACKUP_ID in out | |||
|
|
|||
|
|
|||
| @pytest.mark.skip( | |||
There was a problem hiding this comment.
Can we mark the test with flaky?
You'll need to add
flaky==3.7.0
to requirements-test.txt.
Then mark the test:
@pytest.mark.flaky(max_runs=2, min_passes=1)
But the current timeout seems like 1200 seconds (20 minutes). This is likely problematic with the current test timeout with btlr.
@kurtisvg Is here a way to extend the timeout for a particular test?
I also think the restore operation for a small database should finish quicker. I would rather get annoyed with flaky build cop bot issues until the server side operations are quicker and more consistent.
There was a problem hiding this comment.
@kurtisvg Is here a way to extend the timeout for a particular test?
Nope
There was a problem hiding this comment.
I've used a RetryErrors decorator from the Cloud Spanner test_utils library which will retry the test on DEADLINE_EXCEEDED for a maximum of two attempts. Is this an acceptable alternative to flaky?
|
@larkee Thanks! |
…gleCloudPlatform/python-docs-samples#4641) * test(spanner): fix flaky restore backup sample test * fix: update timeout to match other languages * test: import test_utils and set max retries to 2 Co-authored-by: larkee <larkee@users.noreply.github.com>
Fixes #4544