cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chris Goffinet (JIRA)" <j...@apache.org>
Subject [jira] Updated: (CASSANDRA-404) SSTable write lock not releasing
Date Tue, 01 Sep 2009 04:15:33 GMT

     [ https://issues.apache.org/jira/browse/CASSANDRA-404?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Chris Goffinet updated CASSANDRA-404:
-------------------------------------

    Description: 
Today we saw a node lock up when trying to serve reads. Current investigation shows that some
threads are trying to acquire a ssTable_.lock(). This means that somehow Cassandra has gotten
into a state with a write lock that has not been released (blocking read locks). Thrift connections
started piling up after this occurred.

We are currently investigating the ssTable lock.

  was:
Today we saw a node lock up when trying to serve reads. Current investigation shows that some
threads are trying to acquire a ssTable_.lock(). This means that somehow Cassandra has gotten
into a state with a write lock that has not been released (blocking read locks). Thrift connections
started piling up after this occurred. We also are seeing a race condition with ConcurrentLinkedHashMap
using appendToTail. 

We are currently investigating the ssTable lock. We could remove the ConcurrentLinkedHashMap
for now until that's resolved.


> SSTable write lock not releasing
> --------------------------------
>
>                 Key: CASSANDRA-404
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-404
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>    Affects Versions: 0.4
>            Reporter: Chris Goffinet
>         Attachments: stack.log.gz
>
>
> Today we saw a node lock up when trying to serve reads. Current investigation shows that
some threads are trying to acquire a ssTable_.lock(). This means that somehow Cassandra has
gotten into a state with a write lock that has not been released (blocking read locks). Thrift
connections started piling up after this occurred.
> We are currently investigating the ssTable lock.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message