zhijiangW commented on a change in pull request #11515: [FLINK-16744][task] implement channel state persistence for unaligned checkpoints URL: https://github.com/apache/flink/pull/11515#discussion_r401364009
########## File path: flink-streaming-java/src/main/java/org/apache/flink/streaming/runtime/tasks/SubtaskCheckpointCoordinatorImpl.java ########## @@ -89,17 +101,98 @@ public void checkpointState( checkpointMetaData.getCheckpointId(), checkpointOptions.getTargetLocation()); - CheckpointingOperation.execute( - checkpointMetaData, - checkpointOptions, - checkpointMetrics, - storage, - operatorChain, - taskName, - closeableRegistry, - executorService, - env, - asyncExceptionHandler, - isCanceled); + long startSyncPartNano = System.nanoTime(); + + HashMap<OperatorID, OperatorSnapshotFutures> operatorSnapshotsInProgress = new HashMap<>(operatorChain.getNumberOfOperators()); + try { + for (StreamOperatorWrapper<?, ?> operatorWrapper : operatorChain.getAllOperators(true)) { + StreamOperator<?> op = operatorWrapper.getStreamOperator(); + OperatorSnapshotFutures snapshotInProgress = checkpointStreamOperator( + op, + checkpointMetaData, + checkpointOptions, + storage, + isCanceled); + operatorSnapshotsInProgress.put(op.getOperatorID(), snapshotInProgress); + } + + if (LOG.isDebugEnabled()) { + LOG.debug("Finished synchronous checkpoints for checkpoint {} on task {}", + checkpointMetaData.getCheckpointId(), taskName); + } + + long startAsyncPartNano = System.nanoTime(); + + checkpointMetrics.setSyncDurationMillis((startAsyncPartNano - startSyncPartNano) / 1_000_000); + + // we are transferring ownership over snapshotInProgressList for cleanup to the thread, active on submit + executorService.execute(new AsyncCheckpointRunnable( + operatorSnapshotsInProgress, + checkpointMetaData, + checkpointMetrics, + startAsyncPartNano, + taskName, + closeableRegistry, + env, + asyncExceptionHandler)); + + if (LOG.isDebugEnabled()) { + LOG.debug( + "{} - finished synchronous part of checkpoint {}. Alignment duration: {} ms, snapshot duration {} ms", + taskName, checkpointMetaData.getCheckpointId(), + checkpointMetrics.getAlignmentDurationNanos() / 1_000_000, + checkpointMetrics.getSyncDurationMillis()); + } + } catch (Exception ex) { + // Cleanup to release resources + for (OperatorSnapshotFutures operatorSnapshotResult : operatorSnapshotsInProgress.values()) { + if (null != operatorSnapshotResult) { + try { + operatorSnapshotResult.cancel(); + } catch (Exception e) { + LOG.warn("Could not properly cancel an operator snapshot result.", e); + } + } + } + + if (LOG.isDebugEnabled()) { + LOG.debug( Review comment: I see this log occur twice with the same arguments only different message, and this log actually seems a bit long to impact the normal logics review. I am not sure whether it is worth extracting a separate method for only passing different message in two usages. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services