incubator-cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rob Coli <rc...@digg.com>
Subject Re: Cassandra disk space utilization WAY higher than I would expect
Date Fri, 20 Aug 2010 21:55:12 GMT
On 8/20/10 1:58 PM, Julie wrote:
> Julie<julie.sugar<at>  nextcentury.com>  writes:
>
> Please see previous post but is hinted handoff a factor if the CL is set to ALL?

Your previous post looks like a flush or compaction is causing the node 
to mark its neighbors down. Do you see correlation between memtable 
flushes or compaction and the GMFD marking? Are you running a version of 
Cassandra (<=0.6.4) which has the MESSAGE-DESERIALIZER-POOL bug, and/or 
are you seeing a backup there? That can cause GMFD to incorrectly mark 
nodes down..

https://issues.apache.org/jira/browse/CASSANDRA-1358

However you are correct, I missed that you are writing at CL.ALL. With 
CL.ALL, you should get an exception if nodes are unreachable during a 
write. This means that Hinted Handoff is unlikely to be the culprit for 
your usage bloat. As you clearly have nodes which are occasionally 
considered unreachable.. are you seeing those exceptions?

=Rob

Mime
View raw message