[
https://issues.apache.org/jira/browse/HBASE-13082?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14980263#comment-14980263
]
ramkrishna.s.vasudevan commented on HBASE-13082:
------------------------------------------------
One interesting part while doing this is that of the Secondary replicas. As
per the idea of read replicas the same underlying store file will be having a
reader from the Secondary region always opened. In the current code whenever we
refreshStoreFiles we close the reader that we think is no longer needed. So the
archiving of the file if done by the primary region will be successul. But now
after this change we cannot do that because we need the file to be available
and only the compacted file cleaner can close the reader. Now there are two
case
If suppose the compacted file cleaner has not closed the reader then the
secondary region will have a reader opened on the file and so any external
archive option will fail.
Similarly though the compacted file cleaner will close the reader and try to
archive the compacted file still if there is a reader opened by the secondary
region this archival will fail. Anyway the next cleaner thread will make it
successful. So may be need to see if there are any other implications.
> Coarsen StoreScanner locks to RegionScanner
> -------------------------------------------
>
> Key: HBASE-13082
> URL: https://issues.apache.org/jira/browse/HBASE-13082
> Project: HBase
> Issue Type: Bug
> Reporter: Lars Hofhansl
> Assignee: ramkrishna.s.vasudevan
> Attachments: 13082-test.txt, 13082-v2.txt, 13082-v3.txt,
> 13082-v4.txt, 13082.txt, 13082.txt, HBASE-13082.pdf, HBASE-13082_1_WIP.patch,
> HBASE-13082_2_WIP.patch, HBASE-13082_3.patch, HBASE-13082_4.patch, gc.png,
> gc.png, gc.png, hits.png, next.png, next.png
>
>
> Continuing where HBASE-10015 left of.
> We can avoid locking (and memory fencing) inside StoreScanner by deferring to
> the lock already held by the RegionScanner.
> In tests this shows quite a scan improvement and reduced CPU (the fences make
> the cores wait for memory fetches).
> There are some drawbacks too:
> * All calls to RegionScanner need to be remain synchronized
> * Implementors of coprocessors need to be diligent in following the locking
> contract. For example Phoenix does not lock RegionScanner.nextRaw() and
> required in the documentation (not picking on Phoenix, this one is my fault
> as I told them it's OK)
> * possible starving of flushes and compaction with heavy read load.
> RegionScanner operations would keep getting the locks and the
> flushes/compactions would not be able finalize the set of files.
> I'll have a patch soon.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
