Hi,

Part of the documentation for "taskmanager.memory.managed.consumer-weights"
[1] specifies that "Flink reserves managed memory only if the job contains
managed memory consumers of that type," but if I understand the source code
[2] correctly, all configured weights are always considered. Is
this correct?

Additionally, the OPERATOR consumer weight is used "for built-in
algorithms," but I thought the framework's off-heap fraction was used for
that. Are there more details for this use case?

[1]
https://nightlies.apache.org/flink/flink-docs-release-2.2/docs/deployment/memory/mem_setup_tm/#consumer-weights
[2]
https://github.com/apache/flink/blob/1e6a73616eb4e224a149e684c832f20bcba93148/flink-runtime/src/main/java/org/apache/flink/runtime/util/config/memory/ManagedMemoryUtils.java#L62

Regards,
Alexis.

Reply via email to