[
https://issues.apache.org/jira/browse/HDFS-12348?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16977792#comment-16977792
]
Surendra Singh Lilhore commented on HDFS-12348:
-----------------------------------------------
Our cluster is 7K node cluster and its taken more than one week to upgrade all
the services and applications.
In this one week all the deleted blocks are moved to trash and DN disks are
getting filled. We can have a option to skip trash.
> disable removing blocks to trash while rolling upgrade
> ------------------------------------------------------
>
> Key: HDFS-12348
> URL: https://issues.apache.org/jira/browse/HDFS-12348
> Project: Hadoop HDFS
> Issue Type: New Feature
> Components: datanode
> Reporter: Jiandan Yang
> Assignee: Jiandan Yang
> Priority: Major
> Attachments: HDFS-12348.001.patch, HDFS-12348.002.patch,
> HDFS-12348.003.patch
>
>
> DataNode remove block file and meta file to trash while rolling upgrade，and
> do delete when
> executing finalize.
> This leads disk of datanode to be full, because
> (1) frequently creating and deleting files（eg，Hbase compaction);
> (2) cluster is very big, and rolling upgrade often last several days;
> Current our solution is clean trash by hand, but this is very dangerous in
> product environment.
> we think disable trash of datanode maybe a good method to avoid disk to be
> full.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
