d-c-manning opened a new pull request, #4371:
URL: https://github.com/apache/hbase/pull/4371

   [HBASE-22349](https://issues.apache.org/jira/browse/HBASE-22349)
   
   Slop is a concept used within `SimpleLoadBalancer` but has always been 
referenced in `StochasticLoadBalancer` without any effect. This change proposes 
that we start using the concept of slop, as another trigger for a balancer run. 
This is similar to HBASE-24139, where a server with 0 regions is considered a 
trigger to run the balancer.
   
   In this way we can guarantee balancer attempts to run in other cases that a 
balancer run is desirable, such as only 1 region on a server, or 2x or 3x the 
average region count on a server.
   
   I remove the `normalizeSlop` method for two reasons:
   1. A negative value will indicate the slop value should not be used at all
   2. Values greater than 1 can still be useful (setting slop to 2 would mean 
to trigger a balancer run if a server had >3x the average region count for the 
cluster.)
   
   As far as I can tell, the default slop of `0.001f` never had any meaning for 
the `StochasticLoadBalancer` before this change. I choose to reuse the default 
slop of `0.2f` as used in the `SimpleLoadBalancer`.


-- 
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: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to