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
