[
https://issues.apache.org/jira/browse/CASSANDRA-6796?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13964122#comment-13964122
]
Ben Hood commented on CASSANDRA-6796:
-------------------------------------
Oh, OK - it was hard to tell whether he meant that this issue is fixed by some
other fixes, seeing as there are a couple of other reference tickets earlier on
in the discussion thread, hence my question.
By looking at the assertion at L860 in StorageProxy.java it looks like this
issue is masking an underlying timeout - if this assumption is wrong, please
let me know.
So based on this assumption, what I've done in the meantime is to increase
write_request_timeout_in_ms in cassandra.yaml from 2000 to 10000. My particular
workload is very batchy, so I figured, what could possibly go wrong. For now,
this change in the setting seems to avoid the this issue.
But this might be happenstance - is write_request_timeout_in_ms something I
probably shouldn't be playing with?
> StorageProxy may submit hint to itself (with an assert) for CL.Any
> ------------------------------------------------------------------
>
> Key: CASSANDRA-6796
> URL: https://issues.apache.org/jira/browse/CASSANDRA-6796
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Reporter: Viktor Kuzmin
> Priority: Minor
> Fix For: 2.0.7
>
>
> StorageProxy.mutate may produce WriteTimoutException and with
> ConsistencyLevel.ANY it tries to submitHint. But submitHint function have
> assertation - we may not send hints to ourself. That may lead to exception
> (in case we're among natural endpoints) and hint will be not submitted to
> other endpoints.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
