You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Jonathan Ellis (JIRA)" <ji...@apache.org> on 2010/07/17 04:17:50 UTC
[jira] Created: (CASSANDRA-1293) add locking around row cache
accesses
add locking around row cache accesses
-------------------------------------
Key: CASSANDRA-1293
URL: https://issues.apache.org/jira/browse/CASSANDRA-1293
Project: Cassandra
Issue Type: Bug
Components: Core
Reporter: Jonathan Ellis
Priority: Minor
Fix For: 0.7
CASSANDRA-1267 means we need to lock around removeDeleted on the row cache entry and the write path where we merge in new columns (otherwise there can be a race where we incorrectly continue to remove a column, that has been updated by the writer thread to be newly relevant)
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (CASSANDRA-1293) add locking around row cache
accesses
Posted by "Stu Hood (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CASSANDRA-1293?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12904373#action_12904373 ]
Stu Hood commented on CASSANDRA-1293:
-------------------------------------
How much of a benefit do we get from the row cache being write-through? It would be really nice not having to lock it... or maybe making the locking and write-throughness an option, via CASSANDRA-1158. It would also be easier to manage cache sizes in terms of bytes if we didn't need to deal with mutations to the row cache.
> add locking around row cache accesses
> -------------------------------------
>
> Key: CASSANDRA-1293
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1293
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Reporter: Jonathan Ellis
> Priority: Minor
> Fix For: 0.7.0
>
>
> CASSANDRA-1267 means we need to lock around removeDeleted on the row cache entry and the write path where we merge in new columns (otherwise there can be a race where we incorrectly continue to remove a column, that has been updated by the writer thread to be newly relevant)
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (CASSANDRA-1293) add locking around row cache
accesses
Posted by "Brandon Williams (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CASSANDRA-1293?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12917039#action_12917039 ]
Brandon Williams commented on CASSANDRA-1293:
---------------------------------------------
No quantifiable difference in write speed with or without this patch.
> add locking around row cache accesses
> -------------------------------------
>
> Key: CASSANDRA-1293
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1293
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Reporter: Jonathan Ellis
> Assignee: Jonathan Ellis
> Priority: Minor
> Fix For: 0.7.0
>
> Attachments: 1293.txt
>
>
> CASSANDRA-1267 means we need to lock around removeDeleted on the row cache entry and the write path where we merge in new columns (otherwise there can be a race where we incorrectly continue to remove a column, that has been updated by the writer thread to be newly relevant)
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (CASSANDRA-1293) add locking around row cache
accesses
Posted by "Hudson (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CASSANDRA-1293?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12917192#action_12917192 ]
Hudson commented on CASSANDRA-1293:
-----------------------------------
Integrated in Cassandra #553 (See [https://hudson.apache.org/hudson/job/Cassandra/553/])
lock row cache updates to prevent race condition
patch by jbellis; reviewed by brandonwilliams for CASSANDRA-1293
> add locking around row cache accesses
> -------------------------------------
>
> Key: CASSANDRA-1293
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1293
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Reporter: Jonathan Ellis
> Assignee: Jonathan Ellis
> Priority: Minor
> Fix For: 0.7.0
>
> Attachments: 1293.txt
>
>
> CASSANDRA-1267 means we need to lock around removeDeleted on the row cache entry and the write path where we merge in new columns (otherwise there can be a race where we incorrectly continue to remove a column, that has been updated by the writer thread to be newly relevant)
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (CASSANDRA-1293) add locking around row cache
accesses
Posted by "Jonathan Ellis (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CASSANDRA-1293?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jonathan Ellis updated CASSANDRA-1293:
--------------------------------------
Attachment: 1293.txt
We expect to save [cache hit rate] of a read per write with write-through, since the alternative is to invalidate the cached row at write time and reload it when it's requested again.
Attached patch moves the cache udpate into CFS.apply.
> add locking around row cache accesses
> -------------------------------------
>
> Key: CASSANDRA-1293
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1293
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Reporter: Jonathan Ellis
> Priority: Minor
> Fix For: 0.7.0
>
> Attachments: 1293.txt
>
>
> CASSANDRA-1267 means we need to lock around removeDeleted on the row cache entry and the write path where we merge in new columns (otherwise there can be a race where we incorrectly continue to remove a column, that has been updated by the writer thread to be newly relevant)
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (CASSANDRA-1293) add locking around row cache
accesses
Posted by "Brandon Williams (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CASSANDRA-1293?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12917034#action_12917034 ]
Brandon Williams commented on CASSANDRA-1293:
---------------------------------------------
+1
> add locking around row cache accesses
> -------------------------------------
>
> Key: CASSANDRA-1293
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1293
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Reporter: Jonathan Ellis
> Assignee: Jonathan Ellis
> Priority: Minor
> Fix For: 0.7.0
>
> Attachments: 1293.txt
>
>
> CASSANDRA-1267 means we need to lock around removeDeleted on the row cache entry and the write path where we merge in new columns (otherwise there can be a race where we incorrectly continue to remove a column, that has been updated by the writer thread to be newly relevant)
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.