[
https://issues.apache.org/jira/browse/CASSANDRA-7106?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Victor Igumnov updated CASSANDRA-7106:
--------------------------------------
Description:
See Github for patch:
https://github.com/victori/cassandra
What the patch does:
While I was evaluating Shark/Hive for analyzing Cassandra's data, I could not
get any of the queries to return reliably. Looking further into the issue I
noticed that if just a single fetch for a split fails the whole job is aborted.
So instead of giving up and failing out the whole job, retry fetching the data
a few times before completely giving up.
was:
See Github for patch:
https://github.com/apache/cassandra/pull/34
https://github.com/victori/cassandra
What the patch does:
While I was evaluating Shark/Hive for analyzing Cassandra's data, I could not
get any of the queries to return reliably. Looking further into the issue I
noticed that if just a single fetch for a split fails the whole job is aborted.
So instead of giving up and failing out the whole job, retry fetching the data
a few times before completely giving up.
> Retry multiple times before giving up on fetching splits
> --------------------------------------------------------
>
> Key: CASSANDRA-7106
> URL: https://issues.apache.org/jira/browse/CASSANDRA-7106
> Project: Cassandra
> Issue Type: Improvement
> Components: Hadoop
> Reporter: Victor Igumnov
> Priority: Minor
> Fix For: 2.1 beta1
>
> Attachments: retries-reworked.patch, retry-splits.patch
>
>
> See Github for patch:
> https://github.com/victori/cassandra
> What the patch does:
> While I was evaluating Shark/Hive for analyzing Cassandra's data, I could not
> get any of the queries to return reliably. Looking further into the issue I
> noticed that if just a single fetch for a split fails the whole job is
> aborted. So instead of giving up and failing out the whole job, retry
> fetching the data a few times before completely giving up.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
