Memory load issue

[M.Thomas]

Hi Admins,

I received a alert mail from HPOV yesterday

Memory load, bottleneck situation, current value: 135.01% exceeds configured threshold: 93.00%.

Please tell me where can I find the cause

Thanks in advance

Regards
Thomas.M

[Marco A.]

Hello , this warning tells you that the memory utilization you the system is 135.01% and you setup a threshold at 93.00%.

What you need to do is keep an close eye on your system by using "glance" or "gpm" (graphic glance).
You can stop some un-neccessary processes if memory util is reaching 100%, that in your case is in 135%!!!.

If there is nothing you could stop, you may want to warn your users that the system is very busy, don't start the new processes until some running processes completed.

Hope this helps ,

Regards,

[Patrick Wallek]

I'm not sure how it is calculating 135% usage, unless you have a LOT of paged-out to a swap device.

Check the output of 'swapinfo -tam' and see what you swap device usage is. That may help to explain things.

[M.Thomas]

Thanks Marco & Patrick for your quick response

Now its going fine , but it happened yesterday . I want to know where can I find the cause. Is there any logfile or command to find the cause.

Regards
Thomas.M

[Marco A.]

Actually you need to monitor that a little time, check the swap, a couple of ps -ef , check if the syslog is logging something weird about that. Try to stop unnecessary processes, etc.

rgds

[Bill Hassell · Accepted solution]

There is nothing wrong with your system. 135% memory means that about 35% of the program memory has been moved out to the swap area. That's what Virtual Memory means. All you have to do to use up all your memory is start more programs.

Now it is certainly possible that a program or script is not working as desired and memory is used unnecessarily. While you can find the program that is using the most amount of memory like this:

UNIX95=1 ps -e -o vsz,pid,ruser,args | sort -rn | head -15

But you don't want to fix the problem by killing the biggest memory user without knowing whether it is normal. For instance, the biggest memory user might be oracle and killing that program will make a lot of people angry.

The email message from HP-OV is not very useful. It just says that paging (swapping) is taking place and the system will be running a lot slower than normal. Unless you have someone who can change the applications for you, you can:

1. Learn to live with slow performance
2. Buy a lot more RAM for the computer
3. Tell half of your users to logout and try again later