cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Roman Skvazh (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (CASSANDRA-6698) Many too small SSTables when full repair
Date Wed, 12 Feb 2014 23:21:20 GMT

     [ https://issues.apache.org/jira/browse/CASSANDRA-6698?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Roman Skvazh updated CASSANDRA-6698:
------------------------------------

    Description: 
We have troubles when cassandra drops messages because there is too many (over 10,000 on one
column family) small (from 1Kb to 200Kb, and normal sizes too) and many pending compactions
(over 700).

We are using Leveled compaction with 160Mb sstable size.
PS. Temp fix: stop repair, disable thrift,gossip and wait for compactions to be finished.
Because this, we can not run full repair for about a month :(

  was:
We are have troubles when cassandra drops messages because there is too many (over 10,000
on one column family) small (from 1Kb to 200Kb, and normal sizes too) and many pending compactions
(over 700).

We are using Leveled compaction with 160Mb sstable size.
PS. Temp fix: stop repair, disable thrift,gossip and wait for compactions to be finished.
Because this, we can not run full repair for about a month :(


> Many too small SSTables when full repair
> ----------------------------------------
>
>                 Key: CASSANDRA-6698
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-6698
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: Roman Skvazh
>            Priority: Critical
>         Attachments: cassa-many-small-sstables.txt
>
>
> We have troubles when cassandra drops messages because there is too many (over 10,000
on one column family) small (from 1Kb to 200Kb, and normal sizes too) and many pending compactions
(over 700).
> We are using Leveled compaction with 160Mb sstable size.
> PS. Temp fix: stop repair, disable thrift,gossip and wait for compactions to be finished.
Because this, we can not run full repair for about a month :(



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)

Mime
View raw message