hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Allen Wittenauer (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-2123) Enhancements and Improvement for Host monitor in a big cluster
Date Tue, 06 Nov 2007 07:35:50 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-2123?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12540363
] 

Allen Wittenauer commented on HADOOP-2123:
------------------------------------------

Given that this reads /proc, this patch likely won't work on anything but Linux as written.

> Enhancements and Improvement for Host monitor in a big cluster
> --------------------------------------------------------------
>
>                 Key: HADOOP-2123
>                 URL: https://issues.apache.org/jira/browse/HADOOP-2123
>             Project: Hadoop
>          Issue Type: Improvement
>          Components: dfs, mapred
>    Affects Versions: 0.14.0
>            Reporter: Shuguang Liu
>             Fix For: 0.16.0
>
>         Attachments: hadoop.0.14-pro.tgz, new_commands_v01.patch
>
>
> New Enhancements for HADOOP on big clusters. 
> I put some new commands into HADOOP based on 0.14.0 and modified some bugs especially
in the scheduling part. 
> Enhancements:
> 1. Add Some commands to make it easy to monitor the cluster. 
> hhosts - Display hosts status in the cluster.
> hload - Display host load information including cpu usage, cpu usage and so on.
> hslot - Change host available slot to improve efficiency. 
> hopen - Enable a host on the fly.
> hclose - Disable a host on the fly
> 2. Add Some commands to improve efficiency for a "Big Cluster"
> htop - Bring a job specified to the top prirority on the fly.
> hbot - Bring a job specified to the lowest priority on the fly.
> hjobkill - Kill a job
> hjobs - List job information in detailed format. 
> 3. Enhance the task trunk dispatching method to make it more efficient when a job has
many tasks.
> 4. Improved job submitting efficiency asuring right priority
> [ Show ยป ] Shuguang Liu - 30/Oct/07 01:15 AM New Enhancements for HADOOP on big clusters.
I put some new commands into HADOOP based on 0.14.0 and modified some bugs especially in the
scheduling part. Enhancements: 1. Add Some commands to make it easy to monitor the cluster.
hhosts - Display hosts status in the cluster. hload - Display host load information including
cpu usage, cpu usage and so on. hslot - Change host available slot to improve efficiency.
hopen - Enable a host on the fly. hclose - Disable a host on the fly 2. Add Some commands
to improve efficiency for a "Big Cluster" htop - Bring a job specified to the top prirority
on the fly. hbot - Bring a job specified to the lowest priority on the fly. hjobkill - Kill
a job hjobs - List job information in detailed format. 3. Enhance the task trunk dispatching
method to make it more efficient when a job has many tasks. 4. Improved job submitting efficiency
asuring right priority 

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


Mime
View raw message