[
https://issues.apache.org/jira/browse/FLINK-6537?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16010393#comment-16010393
]
Gyula Fora commented on FLINK-6537:
-----------------------------------
Hi Stefan!
Thank you for all the work :)
I have tested your changes today and some of the issues have been resolved.
Incremental checkpoints now succeed but cannot be restored from them (ether
after cacnellation or under normal failure scenarios)
Savepoints still fail with the previous error. And also there seems to be some
error while discarding subsumed checkpoints. I am attaching the logs containing
the errors:
In the beginning you can see the failed savepoint, then a couple of successful
checkpoints with the warnings and at the end I killed the TM to trigger a
restore which lead to file not found errors in HDFS:
https://gist.github.com/gyfora/2bb5569fb703bbd7e47ba60352f90086
I hope this helps :)
Gyula
> Umbrella issue for fixes to incremental snapshots
> -------------------------------------------------
>
> Key: FLINK-6537
> URL: https://issues.apache.org/jira/browse/FLINK-6537
> Project: Flink
> Issue Type: Bug
> Components: State Backends, Checkpointing
> Affects Versions: 1.3.0
> Reporter: Stefan Richter
> Assignee: Stefan Richter
> Fix For: 1.3.0
>
>
> This issue tracks ongoing fixes in the incremental checkpointing feature for
> the 1.3 release.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
