[ 
https://issues.apache.org/jira/browse/HBASE-6868?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13462319#comment-13462319
 ] 

stack commented on HBASE-6868:
------------------------------

Sorry.  Distracted today.  On DFS_CLIENT_READ_SHORTCIRCUIT_KEY, I'd checked and 
yeah, hadoop 1.0.x has it so patch would be fine for 0.96.

The patch permanently ties the checksumming feature to local short circuit.  Is 
that what we want to do?  Might be ok for 0.94 but we might not want it for 
trunk/0.96 which we want working w/ h2 and hopefully it'll get hdfs-3429 soon.

Can we do this Lars:

{quote}
dfs.client.read.shortcircuit = true
dfs.client.read.shortcircuit.skip.checksum=true
{quote}

It means we read that when we read a WAL with blocks that are local, we'll not 
be checking their checksum.

Should we just turn off the HBASE-5074?  Set 
http://hbase.apache.org/xref/org/apache/hadoop/hbase/regionserver/HRegionServer.html#468
 to false?  Will that get us the old behavior?

[~liulei.cn] Regards "I think there is another problem in local read....I think 
that may lead to RegionServer read wrong data.", I thought it a known hdfs 
issue w/ local read but could not find an issue describing the problem.  I'd 
say file an hdfs issue for it.
                
> Skip checksum is broke; are we double-checksumming by default?
> --------------------------------------------------------------
>
>                 Key: HBASE-6868
>                 URL: https://issues.apache.org/jira/browse/HBASE-6868
>             Project: HBase
>          Issue Type: Bug
>          Components: HFile, wal
>    Affects Versions: 0.94.0, 0.94.1
>            Reporter: LiuLei
>            Priority: Blocker
>             Fix For: 0.94.3, 0.96.0
>
>         Attachments: 6868-0.96-idea.txt, 6868-0.96-v2.txt
>
>
> The HFile contains checksums for decrease the iops, so when Hbase read HFile 
> , that dont't need to read the checksum from meta file of HDFS.  But HLog 
> file of Hbase don't contain the checksum, so when HBase read the HLog, that 
> must read checksum from meta file of HDFS.  We could  add setSkipChecksum per 
> file to hdfs or we could write checksums into WAL if this skip checksum 
> facility is enabled 

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to