[
https://issues.apache.org/jira/browse/HDFS-1103?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12859541#action_12859541
]
Todd Lipcon commented on HDFS-1103:
-----------------------------------
Haven't thought it through entirely, but I think this could be fixed something
like this:
- add a field lastChecksumChunkCorrupt to ReplicaWaitingToBeRecovered,
ReplicaUnderRecovery, and ReplicaRecoveryInfo
- rather than using validateIntegrity to set a shortened block length in
ReplicaWaitingToBeRecovered, set the lastChecksumChunkCorrupt flag
- in DataNode.recoverBlock, use the following logic:
-- If all of the recovery candidates have corrupt last chunks, truncate all of
them like we're doing now
-- If only some of the recovery candidates have corrupt last chunks, don't
allow them to participate in recovery
> Replica recovery doesn't distinguish between flushed-but-corrupted last chunk
> and unflushed last chunk
> ------------------------------------------------------------------------------------------------------
>
> Key: HDFS-1103
> URL: https://issues.apache.org/jira/browse/HDFS-1103
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: data-node
> Affects Versions: 0.21.0, 0.22.0
> Reporter: Todd Lipcon
> Attachments: hdfs-1103-test.txt
>
>
> When the DN creates a replica under recovery, it calls validateIntegrity,
> which truncates the last checksum chunk off of a replica if it is found to be
> invalid. Then when the block recovery process happens, this shortened block
> wins over a longer replica from another node where there was no corruption.
> Thus, if just one of the DNs has an invalid last checksum chunk, data that
> has been sync()ed to other datanodes can be lost.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
