Myasuka 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_r246074413
##########
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:
In my point of view, we should not depend on execution's current status to
determine whether to assign state.
With the given indices, we could figure out the logical plan containing
exact executions needed to be assigned. While try to assign state to all
executions and use execution's status to determine whether assigning initial
state would cause potential bugs indeed.
Moreover, current implementation has already use one base
`restoreLatestCheckpointedState` method to handle both situations to reduce
potential bugs happening. On the other hand, I think
[FLINK-10713](https://issues.apache.org/jira/browse/FLINK-10713) could also
reuse new `restoreLatestCheckpointedState(List<ExecutionVertex>, boolean)`
method.
I'm not sure whether I have expressed my thoughts clearly to convince you,
but please leave any thoughts or concerns if further discussion still needed.
----------------------------------------------------------------
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
