sunhelly commented on a change in pull request #3652:
URL: https://github.com/apache/hbase/pull/3652#discussion_r808716420
##########
File path:
hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/HRegion.java
##########
@@ -253,6 +253,10 @@
public static final String COMPACTION_AFTER_BULKLOAD_ENABLE =
"hbase.compaction.after.bulkload.enable";
+ /** Config for allow split when file count greater than the configured
blocking file count*/
+ public static final String ENABLE_SPLIT_BEFORE_COMPACT_IGNORE_BLOCKING_KEY =
Review comment:
Thanks, @apurtell , I used to consider this issue as a bug. But Duo
advised in the above conversation that this is a feature, and a configure is
needed for this big behavior change, the original words are as follows,
> I think this is a big behavior change? We used to disable split when store
file count is greater than the blocking count, this is feature, not a bug...
>
> So if you want to change this behavior, you'd better write a design doc or
something to better introduce the background, why we should change this
behavior, the advantage and disadvantages, etc.
>
> And if it is not easy to suit all the scenario, is it possible to make it
configurable?
I think split here is somewhat needed to prevent region from endless growing
unexpectedly, I described a scenario in the issue
https://issues.apache.org/jira/browse/HBASE-26242. But it is not required at
all the time as you and Duo said above, and in some scenarios it is suboptimal.
Currently, can we keep it configurable? WDYT? If you agree, I can change the
issue description and clearly the wish for the behavior change. Thanks.
--
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]
