[
https://issues.apache.org/jira/browse/HBASE-12266?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14172546#comment-14172546
]
James Estes commented on HBASE-12266:
-------------------------------------
That isn't helping because as I've mentioned, that guard only assumes that only
one rpc call (the current/next) is the misbehaving one. In my case, it is a
very selective scan, and before the reset, retryAfterOutOfOrderException will
be set to false, but the next rpc call will succeed just fine (because it gets
a tiny amount of data, but not enough to fulfill the scan batch size) and then
retryAfterOutOfOrderException is set back to true.
It really is an endless loop :) I've stopped it after running for over a day.
The logs show it is doing the same loop about every 11s.
My vote here honestly would be to keep the retryAfterOutOfOrderException=true
line, but move the scanTimeout check to be done for all DoNotRetryIOException
(vs just what it is currently checking scanTimeout only for when it is a
UnknownScannerException).
> Slow Scan can cause dead loop in ClientScanner
> -----------------------------------------------
>
> Key: HBASE-12266
> URL: https://issues.apache.org/jira/browse/HBASE-12266
> Project: HBase
> Issue Type: Bug
> Components: Scanners
> Affects Versions: 0.96.0
> Reporter: Qiang Tian
> Priority: Minor
> Attachments: HBASE-12266-master.patch
>
>
> see http://search-hadoop.com/m/DHED45SVsC1.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
