echauchot commented on code in PR #22985:
URL: https://github.com/apache/flink/pull/22985#discussion_r1360310094
##########
flink-runtime/src/main/java/org/apache/flink/runtime/scheduler/adaptive/Executing.java:
##########
@@ -67,13 +77,33 @@ class Executing extends StateWithExecutionGraph implements
ResourceListener {
this.context = context;
Preconditions.checkState(
executionGraph.getState() == JobStatus.RUNNING, "Assuming
running execution graph");
+ this.scalingIntervalMin = scalingIntervalMin;
+ this.scalingIntervalMax = scalingIntervalMax;
+ Preconditions.checkState(
+ !scalingIntervalMin.isNegative(),
+ "{} must be positive integer or 0",
+ JobManagerOptions.SCHEDULER_SCALING_INTERVAL_MIN.key());
+ if (scalingIntervalMax != null) {
+ Preconditions.checkState(
+ scalingIntervalMax.compareTo(scalingIntervalMin) > 0,
+ "{}({}) must be greater than {}({})",
+ JobManagerOptions.SCHEDULER_SCALING_INTERVAL_MAX.key(),
+ scalingIntervalMax,
+ JobManagerOptions.SCHEDULER_SCALING_INTERVAL_MIN.key(),
+ scalingIntervalMin);
+ }
deploy();
// check if new resources have come available in the meantime
context.runIfState(this, this::maybeRescale, Duration.ZERO);
Review Comment:
when the Executing is initially created (without any restart) meaning when
the job transitions from Creating to Executing, what is the point in waiting
the cooldown period ? In the case of a restart, the scalingIntervalMin is
respected because the triggering of the restart (maybeRescale) was not called
before scalingIntervalMin during the life of the pipeline. Then there is no
point in waiting again when we recreate the Executing object (we have already
waited before the restart).
--
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.
To unsubscribe, e-mail: issues-unsubscr...@flink.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
