Dear Experts,

Greetings of the Day.

I'm stuck in a new problem now. I've an assembled intel server having cpu XEON 3.0 GHz, 4 GB RAM DDR, 2 MB Cache, 200 GB x 5 SAS HDD, Intel RAID CARD (RAID 5 implementated). Swap partition's size is 10 GB.
We're running Lotus Domino 7 on server and Lotus notes of clients which is acting as a mail server.

partition details are given below
[root@cms2040 ~]# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/sda11 9.7G 1.6G 7.6G 18% /
/dev/sda1 9.7G 61M 9.1G 1% /boot
none 2.0G 0 2.0G 0% /dev/shm
/dev/sda9 9.7G 1.1G 8.2G 12% /home
/dev/sda2 202G 124G 68G 65% /lotus
/dev/sda3 9.7G 55M 9.1G 1% /opt
/dev/sda8 9.7G 55M 9.1G 1% /tmp
/dev/sda7 9.7G 1.7G 7.5G 18% /usr
/dev/sda5 9.7G 55M 9.1G 1% /usr/local
/dev/sda6 9.7G 109M 9.1G 2% /var

Suddenly a week ago I go through the memory utilization of server. I found it 99% and it is keep holding near to this position for the whole day. I tried to reduce the utilization by restarting the server. But after restarting gradually it increases to 99-100% within 15 mins. I also tried to keep server in normal mode (means don't start the lotus services,i.e. when we login using a "notes" user all the services starts automatically) by not login in to "notes" user.
But the problem still persist. Pls help.

Note: I'm attaching few snapshots for your reference.

Attached Thumbnails

     

I more thing I forgot to mention the OS I'm using that is RHEL 4.5

Also attached the snapshot of TOP command output.

Regards
Arunabh

Attached Thumbnails

 

This seems to be the common question caused by misunderstanding what "free memory" is. If so, there is nothing wrong with the way your system is running. There is plenty of free memory.

In the top output you posted, you have 22288k free and 17996k buffers and 3747348k cached out of a total of 4151404k.

So you are worried because "free" is only half a percent of total.

Memory use by Linux is not that simple. For most practical purposes you should consider most of the memory reported as buffers and cached as also being free. When any program needs memory, most of the cached and buffers memory is available for immediate reuse.

Once you count cached as free, almost all the ram in that top output is free, rather than almost all used.

I am a bit confused about the 1.0g RES for PID 4379. That ought to mean that process is using 1GB of physical ram. But the high value of cached ought to mean that the total physical ram use of all processes combined is far less than 1GB.

I'm not really sure, but my best guess is that process has multiple mappings to the same shared file(s). So up to the 977MB (reported as SHR) of its RES total might represent a smaller amount of ram mapped multiple times. But it would be very strange for a process to have duplicate mappings like that. Maybe it is doing something else strange that makes its RES value appear much larger than the physical ram it is actually using.

Alternately, maybe that RES value is accurate and the cached value is distorted. Pages move pretty freely between cached for the system and RES for a specific process. I didn't think they would be in both places as once. But I'm not sure of that. Or maybe server runs in bursts so top sees non synchronous information caused by top being interrupted by a burst of activity in server. In that case the RES value of server and the cached value of the system could both be correct at nearly the same moment, just not exactly the same moment.

Anyway, all that confusion relates to at most one fourth of your physical ram. Most of your physical ram in that top report is still unambiguously cached and effectively free.

Just small note. As far as I can see on other pictures, "server" process is latest edition of Lotus Notes Server. maybe that helps to better understand it's memory usage.