[
https://issues.apache.org/jira/browse/FLINK-9269?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16458692#comment-16458692
]
ASF GitHub Bot commented on FLINK-9269:
---------------------------------------
Github user StephanEwen commented on the issue:
https://github.com/apache/flink/pull/5934
Concerning serializer snapshots:
- We need to move away from Java Serializing the serializers into the
config snapshots anyways and should do that in the near future.
- I think the config snapshot should be created once when the state is
created, encoded as `byte[]`, and then we only write the bytes. That safes us
from repeated work on every checkpoint and would also prevent concurrent access
to the serializer for creating the snapshot.
> Concurrency problem in HeapKeyedStateBackend when performing checkpoint async
> -----------------------------------------------------------------------------
>
> Key: FLINK-9269
> URL: https://issues.apache.org/jira/browse/FLINK-9269
> Project: Flink
> Issue Type: Bug
> Components: State Backends, Checkpointing
> Affects Versions: 1.5.0
> Reporter: Sihua Zhou
> Assignee: Sihua Zhou
> Priority: Blocker
> Fix For: 1.5.0
>
>
> {code:java}
> @Nonnull
> @Override
> protected SnapshotResult<KeyedStateHandle> performOperation() throws
> Exception {
> // do something
> long[] keyGroupRangeOffsets = new
> long[keyGroupRange.getNumberOfKeyGroups()];
> for (int keyGroupPos = 0; keyGroupPos <
> keyGroupRange.getNumberOfKeyGroups(); ++keyGroupPos) {
> int keyGroupId = keyGroupRange.getKeyGroupId(keyGroupPos);
> keyGroupRangeOffsets[keyGroupPos] = localStream.getPos();
> outView.writeInt(keyGroupId);
> for (Map.Entry<String, StateTable<K, ?, ?>> kvState :
> stateTables.entrySet()) {
> // do something
> }
> }
> // do something
> }
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
