You are viewing a plain text version of this content. The canonical link for it is here.
Posted to hdfs-issues@hadoop.apache.org by "Xiaobing Zhou (JIRA)" <ji...@apache.org> on 2016/10/20 21:51:58 UTC

[jira] [Commented] (HDFS-11038) DiskBalancer: support running multiple commands under one setup of disk balancer

    [ https://issues.apache.org/jira/browse/HDFS-11038?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15593134#comment-15593134 ] 

Xiaobing Zhou commented on HDFS-11038:
--------------------------------------

This can be fixed by closing FileSystem instance maintaine by org.apache.hadoop.hdfs.server.diskbalancer.command. See also  [HDFS-9462.002.patch|https://issues.apache.org/jira/secure/attachment/12834552/HDFS-9462.002.patch].
{code}
110	  /**
111	   * Cleans any resources held by this command.
112	   * <p>
113	   * The main goal is to delete id file created in
114	   * {@link org.apache.hadoop.hdfs.server.balancer.checkAndMarkRunning},
115	   * otherwise, it's not allowed to run multiple commands in a row.
116	   * </p>
117	   */
118	  @Override
119	  public void close() throws IOException {
120	    if (fs != null) {
121	      fs.close();
122	    }
123	  }
124	
{code}


> DiskBalancer: support running multiple commands under one setup of disk balancer
> --------------------------------------------------------------------------------
>
>                 Key: HDFS-11038
>                 URL: https://issues.apache.org/jira/browse/HDFS-11038
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>          Components: balancer & mover
>            Reporter: Xiaobing Zhou
>            Assignee: Xiaobing Zhou
>
> Disk balancer follows/reuses one rule designed by HDFS balancer, that is, only one instance is allowed to run at the same time. This is correct in production system to avoid any inconsistencies, but it's not ideal to write and run unit tests. For example, it should be allowed run plan, execute, scan commands under one setup of disk balancer. One instance rule will throw exception by complaining 'Another instance is running'. In such a case, there's no way to do a full life cycle tests which involves a sequence of commands.



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

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