[GitHub] flink pull request #4870: [FLINK-7783] Don't always remove checkpoints in Zo...

[aljoscha <gi...@git.apache.org>]

GitHub user aljoscha opened a pull request:

    https://github.com/apache/flink/pull/4870

    [FLINK-7783] Don't always remove checkpoints in ZooKeeperCompletedCheckpointStore#recover()

    Alternative version of #4863.
    
    This one actually works. #4863 is not working because I was deserialising checkpoints on demand which is problematic because before checkpoints were registered at the `SharedStateRegistry`. If we deserialise a checkpoint on demand and call dispose on it (as #4863 does) this will potentially remove shared state handles that are needed by the other handles.
    
    This version also fails as soon as one handle cannot be read. If we don't do this, we will break other incremental state handles because we drop their shared state handles. 
    
    R: @StefanRRichter 

You can merge this pull request into a Git repository by running:

    $ git pull https://github.com/aljoscha/flink jira-7783-zookeeper-state-store-fix-simplified2

Alternatively you can review and apply these changes as the patch at:

    https://github.com/apache/flink/pull/4870.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 #4870
    
----
commit 4702bdd96a3baa844850bba47610c2a71ca7f2f1
Author: Aljoscha Krettek <al...@gmail.com>
Date:   2017-10-19T19:26:20Z

    [FLINK-7783] Don't always remove checkpoints in ZooKeeperCompletedCheckpointStore#recover()

----


---

[GitHub] flink pull request #4870: [FLINK-7783] Don't always remove checkpoints in Zo...

[aljoscha <gi...@git.apache.org>]

Github user aljoscha closed the pull request at:

    https://github.com/apache/flink/pull/4870


---