[
https://issues.apache.org/jira/browse/HDFS-17380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17825897#comment-17825897
]
ASF GitHub Bot commented on HDFS-17380:
---------------------------------------
szetszwo commented on PR #6549:
URL: https://github.com/apache/hadoop/pull/6549#issuecomment-1993371108
> When you talk about inaccessible inode, do you mean NameNode unexpected
logic cause some inodes are unreachable?
Yes. The inaccessible inodes may be caused by a bug like HDFS-17045.
> Recover from one earlier checkpoint will not loss data, it will keep both
fsimage and all editlog util the latest transaction.
If it was caused by a bug, replaying the edit log will reproduce to the same
corrupted fsimage.
> ... Will involve to review once understand what it will improve. ...
@Hexiaoqiao, please take a look the current change, which updates the
`FsImageValidation` tool.
> FsImageValidation: remove inaccessible nodes
> --------------------------------------------
>
> Key: HDFS-17380
> URL: https://issues.apache.org/jira/browse/HDFS-17380
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: tools
> Reporter: Tsz-wo Sze
> Assignee: Tsz-wo Sze
> Priority: Major
> Labels: pull-request-available
>
> If a fsimage is corrupted, it may have inaccessible nodes. The
> FsImageValidation tool currently is able to identify the inaccessible nodes
> when validating the INodeMap. This JIRA is to update the tool to remove the
> inaccessible nodes and then save a new fsimage.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
