[GitHub] [hbase] sunhelly edited a comment on pull request #2712: HBASE-25334 TestRSGroupsFallback.testFallback is flaky

[GitBox <gi...@apache.org>]

sunhelly edited a comment on pull request #2712:
URL: https://github.com/apache/hbase/pull/2712#issuecomment-735368143


   > Mind explaining a bit why set a config to 0 could help here?
   
   Yes, it's my pleasure. @Apache9 
   
   In this unit test, it checks if all table regions assigned after balance using `TEST_UTIL.waitUntilAllRegionsAssigned(table)`, and then assert for the RS group of regions.
   But balance() uses aync move, and will throttle move regions. If the waiting time is not longer than the region movement duration, the assertion will be fail.
   When balancer throttle the movement, it will evenly the region movement plans during the `maxBalancingTime`, sending a region movement plan to the AM and sleep a while before another. 
   Set `hbase.balancer.max.balancing=0` will make the throttle of movement when balance be invalid, because throttling only works when `maxBalancingTime>0`. As a result, all the region plans will be be sent to AM without delay, and UT's `TEST_UTIL.waitUntilAllRegionsAssigned(table)` will wait until all the table regions be correctly assigned to its RS group servers.
   
   


----------------------------------------------------------------
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:
users@infra.apache.org