Closed strangemonad-faire closed 4 years ago
Hi Shawn, any more info for reproduce the problem?
Since the Key is UUID, I think it's not easy to be reproduced. Is the issue still can be reproduced from your side?
hi Shawn, any info? If no, here I close for now. Feel free to reopen.
What happened:
pipelines version 0.1.25, gcp
kubeflow
namespaced installPipeline run_details fails to update run details with the following error:
Error while creating or updating run for workflow: 'kubeflow/relationship-graph-updategsdjb-2062-1367216059'. Create error: 'InternalServerError: Failed to store run RUN_NAME to table: Error 1062: Duplicate entry '45474da3-af6e-11e9-86a0-42010a8000e5' for key 'PRIMARY''. Update error: 'Invalid input error: Failed to update run 45474da3-af6e-11e9-86a0-42010a8000e5. Row not found.'
See full trace below. When several runs get in this state, it bogs down the entire mysql pod and ml-pipelines pod
What did you expect to happen:
Updating run data should gracefully handle cases where the data is already reported.
What steps did you take:
I manually deleted the run_details where uuid matched the offending entities in the logs so mysql could become responsive again.
Anything else you would like to add: