StefanRRichter commented on a change in pull request #7009: [FLINK-10712]
Support to restore state when using RestartPipelinedRegionStrategy
URL: https://github.com/apache/flink/pull/7009#discussion_r246329731
##########
File path:
flink-runtime/src/main/java/org/apache/flink/runtime/executiongraph/failover/FailoverRegion.java
##########
@@ -206,13 +206,21 @@ private void restart(long globalModVersionOfFailover) {
try {
if (transitionState(JobStatus.CREATED,
JobStatus.RUNNING)) {
// if we have checkpointed state, reload it
into the executions
- //TODO: checkpoint support restore part
ExecutionVertex cp
- /**
if (executionGraph.getCheckpointCoordinator()
!= null) {
+ // we restart the checkpoint scheduler
for
+ // i) enable new checkpoint could be
triggered without waiting for last checkpoint expired.
+ // ii) ensure the EXACTLY_ONCE
semantics if needed.
+ if
(executionGraph.getCheckpointCoordinator().isPeriodicCheckpointingConfigured())
{
+
executionGraph.getCheckpointCoordinator().stopCheckpointScheduler();
+ }
+
executionGraph.getCheckpointCoordinator().restoreLatestCheckpointedState(
-
connectedExecutionVertexes, false, false);
+
connectedExecutionVertexes, false);
Review comment:
For all cases, as I see it the indexes work as an optimization to just run
the repartitioning for some tasks. I think it would be ok to drop the
precondition check because the assignment is only done once in the very
beginning, so it seems a bit overcautious at this point. If you think that the
optimization is helpful, we can introduce it with the small changes I proposed.
However, I doubt that we see real performance benefits in the reassignment, but
the code is getting more complex, which can make bugs more likely. Maybe you
could just try out if the is an observable performance difference between using
the indexes and just reassigning to all tasks?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
