[
https://issues.apache.org/jira/browse/HBASE-11625?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15260814#comment-15260814
]
Appy commented on HBASE-11625:
------------------------------
Ahhh, the reason for the bug seems quite simple and has always been there in
the description. I guess we all missed it because it's written in a bad way.
Breaking it down.
We check for block magic inside [new HFileBlock(), line
1697|https://github.com/apache/hbase/blob/12f66e3060339acc569ee425385e2dc255bb3e94/hbase-server/src/main/java/org/apache/hadoop/hbase/io/hfile/HFileBlock.java#L1697]
(see the below stack trace) and we verify the checksum after that on line
1705. That's using the data before verifying that it's correct. So if the data
is actually corrupted, we wrongly throw the exception 'invalid hfile block
magic', whereas we should have fallen back to hdfs for getting correct data.
> Reading datablock throws "Invalid HFile block magic" and can not switch to
> hdfs checksum
> -----------------------------------------------------------------------------------------
>
> Key: HBASE-11625
> URL: https://issues.apache.org/jira/browse/HBASE-11625
> Project: HBase
> Issue Type: Bug
> Components: HFile
> Affects Versions: 0.94.21, 0.98.4, 0.98.5, 1.0.1.1, 1.0.3
> Reporter: qian wang
> Assignee: Pankaj Kumar
> Fix For: 2.0.0
>
> Attachments: 2711de1fdf73419d9f8afc6a8b86ce64.gz, HBASE-11625.patch
>
>
> when using hbase checksum,call readBlockDataInternal() in hfileblock.java, it
> could happen file corruption but it only can switch to hdfs checksum
> inputstream till validateBlockChecksum(). If the datablock's header corrupted
> when b = new HFileBlock(),it throws the exception "Invalid HFile block magic"
> and the rpc call fail
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
