[ 
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14978004#comment-14978004
 ] 

Yu Li commented on HBASE-14004:
-------------------------------

{quote}
It's possible that the HDFS sync RPC call fails, but the data is already (may 
partially) transported to the DNs which finally get persisted
{quote}
Does this really happen on your cluster or it's just an assumption 
[~heliangliang] [~chenheng]? 

>From what I could see we will get EOFException when trying to read partially 
>written entries (those larger than 64k and split into multiple packets and 
>some packets got lost due to network or dfs error) and these entries will be 
>abandoned during replication.

If in real case the sync RPC call fails but data somehow persisted, it's more 
like a bug in HDFS or error usage in HBase (like not handling all possible 
exceptions, etc.). I'd suggest to dig deeper into this first before any further 
discussion.

> [Replication] Inconsistency between Memstore and WAL may result in data in 
> remote cluster that is not in the origin
> -------------------------------------------------------------------------------------------------------------------
>
>                 Key: HBASE-14004
>                 URL: https://issues.apache.org/jira/browse/HBASE-14004
>             Project: HBase
>          Issue Type: Bug
>          Components: regionserver
>            Reporter: He Liangliang
>            Priority: Critical
>              Labels: replication, wal
>
> Looks like the current write path can cause inconsistency between 
> memstore/hfile and WAL which cause the slave cluster has more data than the 
> master cluster.
> The simplified write path looks like:
> 1. insert record into Memstore
> 2. write record to WAL
> 3. sync WAL
> 4. rollback Memstore if 3 fails
> It's possible that the HDFS sync RPC call fails, but the data is already  
> (may partially) transported to the DNs which finally get persisted. As a 
> result, the handler will rollback the Memstore and the later flushed HFile 
> will also skip this record.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to