[
https://issues.apache.org/jira/browse/HBASE-14004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15046439#comment-15046439
]
Yu Li commented on HBASE-14004:
-------------------------------
bq. HBase uses NN to do fencing which means HMaster will only consider an RS is
dead when all of its WAL file is closed.
Oh yes, the recover lease logic could make sure of this... Ok, now I agree that
we don't need to persist the acked length on zk, during failover the existing
logic could make sure of consistency. And the to-be-implemented WAL idempotent
feature will make sure that "WAL replay of unacked entry" and "client retry
after RS crash" won't conflict. Thanks for the explanation [~Apache9]!
> [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)
