Details
-
Improvement
-
Status: Closed
-
Minor
-
Resolution: Duplicate
-
1.4.2, 1.5.1, 1.7.2, 1.8.0, 1.9.0
-
None
-
Patch
Description
For people who use docker image and run flink in pods, currently, there is no way to update
MaxDirectMemorySize
(Well one can create a custom version of taskmanager.sh)
As a result, it starts with a value of 8388607T . If the param taskmanager.memory.preallocate is set to false (default) the clean up will only occur when the MaxDirectMemorySize limit is hit and a gc full cycle kicks in. However with pods especially in kuberenete they will get killed because pods do not run at such a high value. (In our case we run 8GB per pod)
The fix would be to allow it be configurable via flink-conf. We can still have a default of 8388607T to avoid a breaking change.
Attachments
Issue Links
- Is contained by
-
FLINK-13980 FLIP-49 Unified Memory Configuration for TaskExecutors
- Resolved
-
FLINK-16614 FLIP-116 Unified Memory Configuration for Job Manager
- Closed
- links to