[
https://issues.apache.org/jira/browse/HDFS-7722?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14351152#comment-14351152
]
Colin Patrick McCabe commented on HDFS-7722:
--------------------------------------------
Eddy and I had an offline discussion about the use of {{Set<File>}} here. It
seems that there is a pervasive assumption elsewhere in the code that
FsVolumeSpi instances are directories. For example, in these interface methods:
{code}
/** @return the base path to the volume */
public String getBasePath();
/** @return the path to the volume */
public String getPath(String bpid) throws IOException;
/** @return the directory for the finalized blocks in the block pool. */
public File getFinalizedDir(String bpid) throws IOException;
{code}
So I think using {{Set<File>}} is OK here for now, since it fits in with the
rest of the code. We will probably have to revisit this later, but it seems
outside the scope of this jira.
One thing I really like about this patch is the fact we no longer hold the
{{FsDatasetImpl}} mutex while scanning every volume. This alone is a very
important improvement.
I think it makes sense to leave the failure information around when removing
volumes due to the disk checker.
{code}
685 LOG.info("Deactivating volumes: " +
686 Joiner.on(",").join(absoluteVolumePaths));
{code}
We should print out the value of {{clearFailure}} here.
+1 once that's addressed.
> DataNode#checkDiskError should also remove Storage when error is found.
> -----------------------------------------------------------------------
>
> Key: HDFS-7722
> URL: https://issues.apache.org/jira/browse/HDFS-7722
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: datanode
> Affects Versions: 2.6.0
> Reporter: Lei (Eddy) Xu
> Assignee: Lei (Eddy) Xu
> Attachments: HDFS-7722.000.patch, HDFS-7722.001.patch,
> HDFS-7722.002.patch
>
>
> When {{DataNode#checkDiskError}} found disk errors, it removes all block
> metadatas from {{FsDatasetImpl}}. However, it does not removed the
> corresponding {{DataStorage}} and {{BlockPoolSliceStorage}}.
> The result is that, we could not directly run {{reconfig}} to hot swap the
> failure disks without changing the configure file.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
