[
https://issues.apache.org/jira/browse/HDFS-2883?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13199075#comment-13199075
]
Arpit Gupta commented on HDFS-2883:
-----------------------------------
The following was noticed.
namenode is running with trash feature enabled.
1. run bin/hadoop dfs -Dfs.trash.interval=0 -rm /path/to/file
the file gets deleted and is not moved to trash (even though trash is on in the
namenode)
2. run bin/hadoop dfs -Dfs.trash.interval=10 -rm /path/to/file
the file gets deleted and is moved to trash
3. now if you run command bin/hadoop dfs -Dfs.trash.interval=1 -rm
/path/to/file
the file gets deleted and moved to trash but is not removed from trash after 1
min as the namenode actually deletes is based on the setting that has been set
when it was started.
So for delete calls the client side configs can be used, but trash clean up is
only server side. The trash settings should be only server side and if the
client wants to skip trash they should only be able to use -skipTrash option.
This was noticed in 1.0 and 0.23
> trash behavior needs to be improved to be consistent so that only server side
> configs are enforced
> --------------------------------------------------------------------------------------------------
>
> Key: HDFS-2883
> URL: https://issues.apache.org/jira/browse/HDFS-2883
> Project: Hadoop HDFS
> Issue Type: Bug
> Affects Versions: 0.24.0, 1.1.0
> Reporter: Arpit Gupta
>
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
