[
https://issues.apache.org/jira/browse/AURORA-900?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14186290#comment-14186290
]
Alex Levenson commented on AURORA-900:
--------------------------------------
Thanks for the quick update.
Fixing this will fix the error reporting, but does propegating a TransientError
mean the client will re-try the request? That would be ideal.
> TransientError response message is not processed correctly
> ----------------------------------------------------------
>
> Key: AURORA-900
> URL: https://issues.apache.org/jira/browse/AURORA-900
> Project: Aurora
> Issue Type: Bug
> Components: Client
> Reporter: Maxim Khutornenko
> Assignee: Maxim Khutornenko
> Priority: Critical
>
> The assembling of error message in case of a TransientError is not done
> correctly in scheduler_client.py.
> Incorrect:
> {noformat}
> raise self.TransientError(", ".join(
> [m for m in resp.details] if resp.details else []))
> {noformat}
> Correct:
> {noformat}
> raise self.TransientError(", ".join(
> [m.message for m in resp.details] if resp.details else []))
> {noformat}
> This is the unfortunate case of python dynamic typing where the unit test did
> not catch the problem due to matching incorrect expectations:
> {noformat}
> mock_thrift_client.killTasks.side_effect = [
> Response(responseCode=ResponseCode.ERROR_TRANSIENT,
> details=["message1", "message2"],
> ...
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
