[ https://issues.apache.org/jira/browse/FLINK-4196?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15375630#comment-15375630 ]
ASF GitHub Bot commented on FLINK-4196: --------------------------------------- GitHub user StephanEwen opened a pull request: https://github.com/apache/flink/pull/2243 [FLINK-4196] [runtime] Remove recovery timestamp from checkpoint restores The 'recoveryTimestamp' was an unsafe wall clock timestamp attached by the master upon recovery. Because this timestamp cannot be relied upon in distributed setups, it is removed here. If we need something like this in the future, we should try and get a globally progress counter or logical timestamp instead. No code in the core Flink repository is affected by this change. You can merge this pull request into a Git repository by running: $ git pull https://github.com/StephanEwen/incubator-flink remove_recovery_timestamp Alternatively you can review and apply these changes as the patch at: https://github.com/apache/flink/pull/2243.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #2243 ---- commit ae65ef4b8ce35aadd089be46b52e7fddb5a3ef85 Author: Stephan Ewen <se...@apache.org> Date: 2016-07-05T08:18:38Z [hotfix] [kafka connector] Minor code cleanups in the Kafka Producer commit c738fcd9f6031becc405c17cc479b9c2340c2414 Author: Stephan Ewen <se...@apache.org> Date: 2016-07-11T18:36:44Z [hotfix] [runtim] Minor code cleanups. commit 116321241923194e9fa6db556681b333197fceed Author: Stephan Ewen <se...@apache.org> Date: 2016-07-13T15:31:35Z [FLINK-4196] [runtime] Remove the 'recoveryTimestamp' from checkpoint restores. The 'recoveryTimestamp' was an unsafe wall clock timestamp attached by the master upon recovery. This this timestamp cannot be relied upon in distributed setups, it is removed. ---- > Remove "recoveryTimestamp" > -------------------------- > > Key: FLINK-4196 > URL: https://issues.apache.org/jira/browse/FLINK-4196 > Project: Flink > Issue Type: Bug > Components: State Backends, Checkpointing > Affects Versions: 1.0.3 > Reporter: Stephan Ewen > Assignee: Stephan Ewen > > I think we should remove the {{recoveryTimestamp}} that is attached on state > restore calls. > Given that this is a wall clock timestamp from a master node, which may > change when clocks are adjusted, and between different master nodes during > leader change, this is an unsafe concept. -- This message was sent by Atlassian JIRA (v6.3.4#6332)