[ 
https://issues.apache.org/jira/browse/HBASE-4583?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13144580#comment-13144580
 ] 

Lars Hofhansl commented on HBASE-4583:
--------------------------------------

I don't think that HBASE-2856 will address the issues I raised above (which is 
caused by deleting KVs from the memstore).

Updating the WAL first, then applying the changes to the memstore, and making 
them visible immediately, all while holding the rowlock is a relatively little 
change. It'll make Increment/Append/ICV behave correctly in the face of failing 
WAL sync's, but will also significantly slow them down (in terms of throughput, 
but not latency).

I will factor some common bits out (similar to my initial patch), so that if we 
want to tackle this in the future (for example by maintaining the lowest 
current readpoint in use, or something), there'll be just one place to change 
it.

If one want performance, one would set writeToWal to false; but maybe these 
three ops should get another flag to allow for delayed log flush (which would 
be the current behavior).

Lastly, we can just document that the Append and Increment operations are 
convenience methods for grouping many updates into a single roundtrip (not to 
achieve consistency between CFs).

                
> Integrate RWCC with Append and Increment operations
> ---------------------------------------------------
>
>                 Key: HBASE-4583
>                 URL: https://issues.apache.org/jira/browse/HBASE-4583
>             Project: HBase
>          Issue Type: Bug
>            Reporter: Lars Hofhansl
>            Assignee: Lars Hofhansl
>             Fix For: 0.94.0
>
>         Attachments: 4583-v2.txt, 4583-v3.txt, 4583-v4.txt, 4583.txt
>
>
> Currently Increment and Append operations do not work with RWCC and hence a 
> client could see the results of multiple such operation mixed in the same 
> Get/Scan.
> The semantics might be a bit more interesting here as upsert adds and removes 
> to and from the memstore.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: 
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Reply via email to