hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jim Kellerman (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-2135) [hbase] region 'requests' attribute is unreliable
Date Tue, 06 Nov 2007 21:16:51 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-2135?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12540573

Jim Kellerman commented on HADOOP-2135:

Just remember that the number of requests is per heart-beat interval and is not averaged or
accumulated in any way.

Depending on when you look, the rate could either be very high or very low.

We need a better way of doing this, when we get more serious about load balancing.

> [hbase] region 'requests' attribute is unreliable
> -------------------------------------------------
>                 Key: HADOOP-2135
>                 URL: https://issues.apache.org/jira/browse/HADOOP-2135
>             Project: Hadoop
>          Issue Type: Bug
>          Components: contrib/hbase
>            Reporter: stack
>            Priority: Minor
> Watching the hbase webUI, the request counter in the load column reported in the list
of regionservers doesn't make sense.  Refreshing, the request count will tend to rise for
a while as you'd expect but then they'll start to fall and will often drop to zero.  This
is on regionservers with constant count of regions (I'd expect load to be zero on a new region
after a split).

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message