[ 
https://issues.apache.org/jira/browse/HBASE-6298?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13405143#comment-13405143
 ] 

Zhihong Ted Yu edited comment on HBASE-6298 at 7/2/12 5:25 PM:
---------------------------------------------------------------

This issue should be asked on dev mailing list first.
There is a sloppy factor used by load balancer (defaulting to 20%):
{code}
  public void setConf(Configuration conf) {
    this.slop = conf.getFloat("hbase.regions.slop", (float) 0.2);
{code}
The log message cited above came from this check:
{code}
    int floor = (int) Math.floor(average * (1 - slop));
    int ceiling = (int) Math.ceil(average * (1 + slop));
    if (serversByLoad.lastKey().getLoad() <= ceiling &&
       serversByLoad.firstKey().getLoad() >= floor) {
{code}

The average number of regions is 37 for Dave's cluster.
(44-37)/37=19%

@Dave:
I suggest you tighten "hbase.regions.slop"
                
      was (Author: zhi...@ebaysf.com):
    This issue should be asked on dev mailing list first.
There is a sloppy factor used by load balancer (defaulting to 20%):
{code}
  public void setConf(Configuration conf) {
    this.slop = conf.getFloat("hbase.regions.slop", (float) 0.2);
{code}
The average number of regions is 37 for Dave's cluster.
(44-37)/37=19%

@Dave:
I suggest you tighten "hbase.regions.slop"
                  
> Region balancer not balancing
> -----------------------------
>
>                 Key: HBASE-6298
>                 URL: https://issues.apache.org/jira/browse/HBASE-6298
>             Project: HBase
>          Issue Type: Bug
>          Components: master
>    Affects Versions: 0.94.0
>            Reporter: Dave Revell
>         Attachments: master_startup.log
>
>
> Despite regions being unbalanced, the load balancer takes no action. On my 
> cluster the least-loaded regionserver has 33 regions and the most-loaded 
> regionserver has 44 regions. My cluster has 1084 regions and 29 servers. It 
> might be relevant that a 30th server used to belong to the cluster but was 
> removed.
> The master log has some strange entries when the balancer runs. The attached 
> log file was generated by restarting the master, then running "balancer" in 
> the shell.

--
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

        

Reply via email to