[
https://issues.apache.org/jira/browse/HBASE-7672?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13562974#comment-13562974
]
Sergey Shelukhin commented on HBASE-7672:
-----------------------------------------
So this basically discards the compaction request and makes the store
"recalculate" one?
Is it safe to call finishRequest on incomplete request from queue?
Also, if the request is dequeued and discarded anyway, what's the point of
putting it there in the first place?
Maybe it can be generated once before compaction.
bq. existedRequest = ((PriorityCompactionQueue) smallCompactions.getQueue())
1) "existingRequest"
2) Casts look hacky. If we know the type why not use the type.
Also, PriorityCompactionQueue is not actually used as a field anywhere in the
patch.
bq. Compcation
typo
> Merging compaction requests in the queue for same store
> -------------------------------------------------------
>
> Key: HBASE-7672
> URL: https://issues.apache.org/jira/browse/HBASE-7672
> Project: HBase
> Issue Type: Improvement
> Components: Compaction
> Affects Versions: 0.94.4
> Reporter: chunhui shen
> Assignee: chunhui shen
> Fix For: 0.96.0
>
> Attachments: HBASE-7672.patch
>
>
> With a high write presesure, we could found many compaction requests for same
> store in the compaction queue.
> I think we could merge compaction requests for same store to increase
> compaction efficiency greately. It is so in 0.90 version because doing
> compacting files selection only when executing compaction
> e.g.
> {code}
> SmallCompation active count:1,Queue:
> regionName=abctest,90F9AUIPK4YO47W55WS4R8RSKGDFNRYBNB79COYKHNQD9F62G7,1359104485823.f05568c159940b8a72bd84c988388ad3.,
> storeName=c1, fileCount=4, fileSize=371.1m (212.0m, 53.0m, 53.0m, 53.0m),
> priority=15, time=56843340270506608
> regionName=abctest,90F9AUIPK4YO47W55WS4R8RSKGDFNRYBNB79COYKHNQD9F62G7,1359104485823.f05568c159940b8a72bd84c988388ad3.,
> storeName=c1, fileCount=4, fileSize=330.4m (171.3m, 53.0m, 53.0m, 53.0m),
> priority=11, time=56843401092063608
> {code}
> We could merge these two compaction requests
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
