[
https://issues.apache.org/jira/browse/CASSANDRA-2388?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13057461#comment-13057461
]
Brandon Williams commented on CASSANDRA-2388:
---------------------------------------------
{quote}
This is making the presumption that the hadoop cluster is only used with CFIF.
The TT could still be useful for other jobs submitted.
{quote}
I'm fine with that assumption. If you want to run other jobs, use a different
cluster. Cassandra's JVM is eating wasteful memory at that point.
{quote}
Furthermore a blacklisted TT does't automatically come back - it needs to be
manually restarted. Isn't this creating more headache for operations?
{quote}
I don't think this is actually the case, see HADOOP-4305
{quote}
I dont think we should require the TT to be running locally. The whole idea is
to support access to Cassandra data from hadoop even if it's just an import.
This patch does spend a lot of time dealing with non local data for that reason.
{quote}
I'm fine with dropping support for non-colocated TTs, or at least saying
there's no DC-specific support. Because frankly, that is a very suboptimal
thing to do, transfer the data across the network all the time, and flies in
the face of Hadoop's core principles.
> ColumnFamilyRecordReader fails for a given split because a host is down, even
> if records could reasonably be read from other replica.
> -------------------------------------------------------------------------------------------------------------------------------------
>
> Key: CASSANDRA-2388
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2388
> Project: Cassandra
> Issue Type: Bug
> Components: Hadoop
> Affects Versions: 0.7.6, 0.8.0
> Reporter: Eldon Stegall
> Assignee: Jeremy Hanna
> Labels: hadoop, inputformat
> Fix For: 0.7.7, 0.8.2
>
> Attachments: 0002_On_TException_try_next_split.patch,
> CASSANDRA-2388-addition1.patch, CASSANDRA-2388.patch, CASSANDRA-2388.patch,
> CASSANDRA-2388.patch
>
>
> ColumnFamilyRecordReader only tries the first location for a given split. We
> should try multiple locations for a given split.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
