[jira] Commented: (HADOOP-5133) FSNameSystem#addStoredBlock does not handle inconsistent block length correctly

["Konstantin Shvachko (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HADOOP-5133?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12678989#action_12678989 ] 

Konstantin Shvachko commented on HADOOP-5133:
---------------------------------------------

This patch definitely reduces the probability "committing" an incorrect block. Choosing longer block out of all replicas is better than selecting a shorter one.
But this does not answer the question what if (as a result of a software bug or unfortunate circumstance) the longest block is actually a wrong replica. In general the name-node does not have a definite criteria to judge which replica is the right and which is not except for the generation stamp. And it would be wrong to *silently* make such a decision based on the size (or any other artificial convention).
I am coming to a conclusion that the honest way to deal with this is to declare all replicas corrupt in this case, that is when this is the last block of a file being written to. This will be reported in fsck and an administrator or the user can deal with it.
This should happen only as a result of an error in the code so may be we should just treat it as a corruption.

> FSNameSystem#addStoredBlock does not handle inconsistent block length correctly
> -------------------------------------------------------------------------------
>
>                 Key: HADOOP-5133
>                 URL: https://issues.apache.org/jira/browse/HADOOP-5133
>             Project: Hadoop Core
>          Issue Type: Bug
>          Components: dfs
>    Affects Versions: 0.18.2
>            Reporter: Hairong Kuang
>            Assignee: Hairong Kuang
>             Fix For: 0.20.0
>
>         Attachments: inconsistentLen.patch, inconsistentLen1.patch, inconsistentLen2.patch
>
>
> Currently NameNode treats either the new replica or existing replicas as corrupt if the new replica's length is inconsistent with NN recorded block length. The correct behavior should be
> 1. For a block that is not under construction, the new replica should be marked as corrupt if its length is inconsistent (no matter shorter or longer) with the NN recorded block length;
> 2. For an under construction block, if the new replica's length is shorter than the NN recorded block length, the new replica could be marked as corrupt; if the new replica's length is longer, NN should update its recorded block length. But it should not mark existing replicas as corrupt. This is because NN recorded length for an under construction block does not accurately match the block length on datanode disk. NN should not judge an under construction replica to be corrupt by looking at the inaccurate information:  its recorded block length.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.