[jira] [Commented] (SPARK-8688) Hadoop Configuration has to disable client cache when writing or reading delegation tokens.

["Steve Loughran (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SPARK-8688?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15130421#comment-15130421 ] 

Steve Loughran commented on SPARK-8688:
---------------------------------------

Has anyone filed a bug against HDFS for this?

> Hadoop Configuration has to disable client cache when writing or reading delegation tokens.
> -------------------------------------------------------------------------------------------
>
>                 Key: SPARK-8688
>                 URL: https://issues.apache.org/jira/browse/SPARK-8688
>             Project: Spark
>          Issue Type: Bug
>          Components: YARN
>    Affects Versions: 1.5.0
>            Reporter: SaintBacchus
>            Assignee: SaintBacchus
>             Fix For: 1.5.0
>
>
> In class *AMDelegationTokenRenewer* and *ExecutorDelegationTokenUpdater*, Spark will write and read the credentials.
> But if we don't disable the *fs.hdfs.impl.disable.cache*, Spark will use cached  FileSystem (which will use old token ) to  upload or download file.
> Then when the old token is expired, it can't gain the auth to get/put the hdfs.
> (I only tested in a very short time with the configuration:
> dfs.namenode.delegation.token.renew-interval=3min
> dfs.namenode.delegation.token.max-lifetime=10min
> I'm not sure whatever it matters.
>  )



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

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org