[
https://issues.apache.org/jira/browse/FLINK-3322?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15186966#comment-15186966
]
ASF GitHub Bot commented on FLINK-3322:
---------------------------------------
Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1769#issuecomment-194250033
I think the right way to solve that would actually not be in the
MemoryManager (to cache), but in the operators that know that they will need
the memory again and again and will hold onto it.
For another feature in the code, I am currently changing the access to
memory such that it goes through "allocators". These are used by for example
the sorter, to get their memory segments from the memory manager. It would be
simple to create them in the batch operators such that they don't immediately
release to the mem manager, but only after the operator is disposed. That means
memory segments will be held onto through all iterations.
> MemoryManager creates too much GC pressure with iterative jobs
> --------------------------------------------------------------
>
> Key: FLINK-3322
> URL: https://issues.apache.org/jira/browse/FLINK-3322
> Project: Flink
> Issue Type: Bug
> Components: Distributed Runtime
> Affects Versions: 1.0.0
> Reporter: Gabor Gevay
> Assignee: Gabor Horvath
> Priority: Critical
> Fix For: 1.0.0
>
>
> When taskmanager.memory.preallocate is false (the default), released memory
> segments are not added to a pool, but the GC is expected to take care of
> them. This puts too much pressure on the GC with iterative jobs, where the
> operators reallocate all memory at every superstep.
> See the following discussion on the mailing list:
> http://apache-flink-mailing-list-archive.1008284.n3.nabble.com/Memory-manager-behavior-in-iterative-jobs-tt10066.html
> Reproducing the issue:
> https://github.com/ggevay/flink/tree/MemoryManager-crazy-gc
> The class to start is malom.Solver. If you increase the memory given to the
> JVM from 1 to 50 GB, performance gradually degrades by more than 10 times.
> (It will generate some lookuptables to /tmp on first run for a few minutes.)
> (I think the slowdown might also depend somewhat on
> taskmanager.memory.fraction, because more unused non-managed memory results
> in rarer GCs.)
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
