holdenk opened a new pull request #28614:
URL: https://github.com/apache/spark/pull/28614
### What changes were proposed in this pull request?
Increase the timeout and register the listener earlier to avoid any race
condition of the job starting before the listener is registered.
### Why are the changes needed?
The test is currently semi-flaky.
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
I'm currently running the following bash script on my dev machine to verify
the flakiness decreases. It has gotten to 356 iterations without any test
failures so I believe issue is fixed.
```
set -ex
./build/sbt clean compile package
((failures=0))
for (( i=0;i<1000;++i )); do
echo "Run $i"
((failed=0))
./build/sbt "core/testOnly
org.apache.spark.scheduler.WorkerDecommissionSuite" || ((failed=1))
echo "Resulted in $failed"
((failures=failures+failed))
echo "Current status is failures: $failures out of $i runs"
done
```
----------------------------------------------------------------
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:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
