[ 
https://issues.apache.org/jira/browse/HDFS-10770?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Junping Du updated HDFS-10770:
------------------------------
    Target Version/s:   (was: 2.8.0)

> Namenode shouldn't change storage info if client reports different storage. 
> (via reportBadBlock)
> ------------------------------------------------------------------------------------------------
>
>                 Key: HDFS-10770
>                 URL: https://issues.apache.org/jira/browse/HDFS-10770
>             Project: Hadoop HDFS
>          Issue Type: Bug
>            Reporter: Rushabh S Shah
>
> From HDFS-10348 [comment | 
> https://issues.apache.org/jira/browse/HDFS-10348?focusedCommentId=15271522&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-15271522].
> This jira is created to discuss whether we should change the storageInfo in 
> triplets if the client reports bad block with storage which is different than 
> what the namenode thinks.
> Here is the relevant code:
> {code:title=DatanodeStorageInfo.java|borderStyle=solid}
> public AddBlockResult addBlock(BlockInfo b, Block reportedBlock) {
>     // First check whether the block belongs to a different storage
>     // on the same DN.
>     AddBlockResult result = AddBlockResult.ADDED;
>     DatanodeStorageInfo otherStorage =
>         b.findStorageInfo(getDatanodeDescriptor());
>     if (otherStorage != null) {
>       if (otherStorage != this) {
>         // The block belongs to a different storage. Remove it first.
>         otherStorage.removeBlock(b);
>         result = AddBlockResult.REPLACED;
>       } else {
>         // The block is already associated with this storage.
>         return AddBlockResult.ALREADY_EXIST;
>       }
>     }
>     b.addStorage(this, reportedBlock);
>     blocks.add(b);
>     return result;
>   }
> {code}



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

---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org

Reply via email to