hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Thomas Scott (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-11721) When Standby NameNode is paused using kill -SIGSTOP clients hang rather than moving to the active.
Date Tue, 02 May 2017 20:20:04 GMT

    [ https://issues.apache.org/jira/browse/HDFS-11721?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15993655#comment-15993655
] 

Thomas Scott commented on HDFS-11721:
-------------------------------------

[~yzhangal] from HADOOP-12672 it seems the symptom is an exception thrown by the client? I
see no exception, in fact the client continues successfully as soon as the SBNN is unpaused.
I have reproduced on CDH5.8.2 as well.

> When Standby NameNode is paused using kill -SIGSTOP clients hang rather than moving to
the active.
> --------------------------------------------------------------------------------------------------
>
>                 Key: HDFS-11721
>                 URL: https://issues.apache.org/jira/browse/HDFS-11721
>             Project: Hadoop HDFS
>          Issue Type: Bug
>          Components: hdfs-client
>            Reporter: Thomas Scott
>            Priority: Minor
>
> Using kill -SIGSTOP on the standby namenode causes clients to hang rather than moving
to the active. To reproduce:
> 1. Run kill -SIGSTOP <SBNN pid>
> 2. Run hdfs dfs -ls / (this may not show the issue as it may try the active namenode
first. If this happens, failover the NameNodes and the issue will occur)
> 3. To force the issue run hdfs dfs -ls hdfs://<sbnn ip>:8020/ 
> This causes the client to hang with no timeout until the SBNN is resumed.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

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


Mime
View raw message