[
https://issues.apache.org/jira/browse/HBASE-7070?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
chunhui shen updated HBASE-7070:
--------------------------------
Attachment: HBASE-7070v3.patch
Attaching patchV3
bq.We have enough retrying and loops of retries already in the client
Retry in ClientScanner is different with ServerCallable#withRetries, the former
will reset scanner and the latter only send next request again.
bq.Should we retry if OutOfOrderException only, and retry once only rather than
some maximum retries number?
I think retrying once is same with some maximum retries number, or else we
don't do the retry for OutOfOrderException..
@ted
2 * CLIENT_RETRIES_NUMBER is only used to simulate the infinite retry number,
so we expect maximal try number = CLIENT_RETRIES_NUMBER, and I has amended it
in assertTrue from RegionServerWithScanTimeout.tryNumber <=
2*CLIENT_RETRIES_NUMBER to RegionServerWithScanTimeout.tryNumber <=
CLIENT_RETRIES_NUMBER
> Scanner may retry forever after HBASE-5974
> -------------------------------------------
>
> Key: HBASE-7070
> URL: https://issues.apache.org/jira/browse/HBASE-7070
> Project: HBase
> Issue Type: Bug
> Components: Client
> Affects Versions: 0.96.0
> Reporter: chunhui shen
> Assignee: chunhui shen
> Fix For: 0.96.0
>
> Attachments: HBASE-7070.patch, HBASE-7070v2.patch, HBASE-7070v3.patch
>
>
> After the patch of HBASE-5974
> Suppose The process is as the following:
> 1.A next request is very large, so first time it is failed because of timeout
> 2.The second time it is failed again because of
> CallSequenceOutOfOrderException
> 3.We will retry this next request again after reset scanner, so it is also
> very large and failed again because of timeout
> 4.CallSequenceOutOfOrderException again
> 5.Repeated the above forever...
> See the comment in HBASE-5974 for more
> https://issues.apache.org/jira/browse/HBASE-5974?focusedCommentId=13486658&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13486658
--
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
