wchevreuil commented on PR #4799: URL: https://github.com/apache/hbase/pull/4799#issuecomment-1448139983
> An update on testing. In addition to adding a unit test which considers different combinations of prefetch ratio on the currently hosting region server as well as historical prefetch ratio, these changes were tested on a real cluster. However, it was only tested for happy condition viz. if the prefetch ratio on a region server was good, then the region was not moved back to the old region server even though it was hosted and prefetched on the older region server in the part. This test was done by disabling all other cost functions by setting their multiplier values to 0. Every cost function provides a parameter which can be set to 0 to disable that particular cost function. Thanks for the updates! I was looking at the existing functions and based on the respective multipliers default values, I guess region skewness and locality are currently considered the most relevant factors for StochasticLoadBalancer. The prefetch ratio function would be most relevant for non-hdfs deployments, which then makes locality irrelevant. In those cases, prefetch and region skewness would be competing to prevail over the final balancer decision. We should also document this and mention that the behaviour is configurable via the function multiplier property. -- 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...@hbase.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
