Roman comments

Results 21 comments of


                                            Roman

[FLINK-27570][runtime] Fix checkFailureCounter in CheckpointFailureManager for finalize failure

Thanks a lot for your analysis @gaoyunhaii! I think we can distinguish these cases in different ways: 1. [`checkNoPartlyOperatorsFinishedVertexUsedUnionListState`](https://github.com/apache/flink/blob/74f90d722f7be5db5298b84626935a585391f0df/flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/DefaultCheckpointPlan.java#L186) returns a result value (e.g. `boolean`) which then is checked in...

[WIP][FLINK-26372][runtime][state] Allow to configure Changelog Storage per program

Superseded by #21637.

[FLINK-31238] [WIP] Use IngestDB to speed up Rocksdb rescaling recovery (rebased)

One more thing, we should probably update the UI and show the new flag under `Checkpoints / Configuration`. But that can be a separate PR.

[FLINK-34417] Log Job ID via MDC

@flinkbot run azure

[FLINK-34417] Log Job ID via MDC

@flinkbot run azure

[FLINK-34417] Log Job ID via MDC

@flinkbot run azure

[FLINK-34417] Log Job ID via MDC

The failure is unrelated: FLINK-34420

[FLINK-34417] Log Job ID via MDC

@flinkbot run azure

[FLINK-34417] Log Job ID via MDC

Re-ran benchmarks - no regressions found: [jmh-result-runtime.csv](https://github.com/apache/flink/files/14537980/jmh-result-runtime.csv) [jmh-result-scheduler.csv](https://github.com/apache/flink/files/14537982/jmh-result-scheduler.csv)

[FLINK-34199] Add tracing for durations of rescaling/restoring RocksDB incremental checkpoints from downloaded and local state

Right :sweat_smile: Then I guess the only confusion for me is about "local" in names and docs: remote state that is downloaded can be called "local", but it's pure implementation...