How to troubleshoot High Load

[vmlover]

This is a greate article so I thought I ll share

http://www.linuxjournal.com/magazine...-load?page=0,1

[emad]

Incase the load is due to I/O then you can use iotop to check which application has high reads and writes

On centos you can install it as per the following

yum install iotop

[mike2011]

Hi
Its easy to reduce the load, 1st find the processes which consume high resources, then either kill them or terminate them.
top
ps -a

will help to find the processes.

kill or term will help to end it.

[vmlover]

Yes you can look to see if any of the processes that are causing high load, are something that can be done without. Or you can try to adjust those processes so that they do not require such a high level of load. Sometimes you just need to make adjustments and then test those adjustments until you find the ideal solution.

[Sadurni]

My internet speed has been creating problem now a days, My friend suggest me to remove or kill Trouble shoot from your computer, I have no idea about this, Let me know, what is trouble shoot? How can I remove this, there is any error in my system due his..

[jacobwallace]

Hi
find the processes which consume high resources, then either kill them or terminate them.i think this is the simple way to reduce the load.

[thecommis]

This is good article on datacenter talk tools. Keep it up.

[truedon]

great advice Check the top, iotop, memstat and disk utilization with iostat.

A huge problem in debugging a running system is that you may not catch the problem real-time so setting up proper graphing for log analysis and so forth is required, then a further configuration of range based alarms and triggers to ensure the system can run on auto-pilot is optimal. Using munin and monitd with nagios it is possible to achieve this, also cactus is good for resource graphing. I often have it running on one monitor whilst I work incase anything happens on the Disk IO level of TCP/IP. Graphing is the easiest way to forecast, see problems and debug.

[CoolGeek]

Originally Posted by truedon

great advice Check the top, iotop, memstat and disk utilization with iostat.

A huge problem in debugging a running system is that you may not catch the problem real-time so setting up proper graphing for log analysis and so forth is required, then a further configuration of range based alarms and triggers to ensure the system can run on auto-pilot is optimal. Using munin and monitd with nagios it is possible to achieve this, also cactus is good for resource graphing. I often have it running on one monitor whilst I work incase anything happens on the Disk IO level of TCP/IP. Graphing is the easiest way to forecast, see problems and debug.

So true! You really must graph out everything to see how your system performs and the peaks effect on your service. Mostly all sites I have been brought into where having downtime issues only during their peak, so find a budget solution to handle the peak has been a good use of Cloud service providers and automatic provisioning with Chef and Vagrant.