Memory taking 95% for X server.
 

Greetings

I have a Machine installed with RHEL 4.7.

The machine gets freeze every day and by using the top command i get X taking 95% of the memory.

I have 8GB of RAM.
localhost % free -m
total used free shared buffers
Mem: 7984 7941 43 0 5
-/+ buffers/cache: 7875 109
Swap: 16347 16347 0

Localhost % cat /proc/driver/nvidia/cards/0
Model: Quadro FX 5600
IRQ: 185
Video BIOS: 60.80.13.00.01
Card Type: PCI-E
DMA Size: 40 bits
DMA Mask: 0xffffffffff
Bus Location: 02.00.0


Localhost % cat /proc/driver/nvidia/version
NVRM version: NVIDIA UNIX x86_64 Kernel Module 190.42 Tue Oct 20 20:25:42 PDT 2009
GCC version: gcc version 3.4.6 20060404 (Red Hat 3.4.6-10)


Top command output
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
20244 root 15 0 24.1g 7.4g 1944 S 0.3 95.0 112:57.72 X


PLease help me out in trouble shooting this problem.

If need further please feel free to contact.


Regards

Hard freezes are a symptom of either a buggy kernel module or a hardware problem.

I don't know if you have some problem or not with an application leaking ram, because the output you pasted is so messy (please, use code tags next time to wrap the terminal output). But in any case, that could (at most) activate the kernel OOM killer, that would kill the offending app (hopefully, but it might be any random one as well). Filling your ram doesn't freeze your computer.

About kernel modules, I'd check first any 3rd party module that's not part of the kernel (nvidia is an example of that, so try to reproduce your problem using the vesa driver or the nv one).

About hardware, check your ram using memtest (most livecds include a boot option to launch it) and watch over the temperature of your cpu, video card and HDs. Check also your HDs with smartmontools.

ps. Filling your ram doesn't crash your pc, but if you start to swap heavily to disk then it will turn really slow, to the point that it might seem to be frozen. However, the HD should show some heavy activity when that happens.

The following could be one of your problems.

http://us.download.nvidia.com/XFree8...hapter-10.html

Another problem could be some Xorg versions have a memory leak or pixmaps causing the memory leak. The only way to fix this is set a limit for pixmaps or upgrade Xorg to at least 7.2 or higher.

I suggest use CentOS, so you are not confined to buying when you need to upgrade.

I doubt nVidia states their drivers being stable for every distribution although they are a reputable company that makes good drivers. I use Gentoo portage tree to find out which versions of the desire program are stable and reliable for a whole setup. At this time nVidia drivers version 180.60 are stated as stable while 190.42 is not, so this is something to think about. Even though this is for Gentoo, so you could this information for other distributions..


On my setup X Window Server only uses around 14 megabytes of RAM. I got this when using pmap -d pid as root. The writable or private is the value that I used.

Upgrading to a newer version may be less effort than diagnosing this problem.

The output was messy but it also unambiguously showed that there is a memory leak somewhere in the X process. Not a hardware problem, not a buggy kernel module.

Doesn't look to me like it fits at all.

I don't know about any of that, but it fits the symptoms. By the "pixmaps" part of that suggestion do you still mean a bug in X or do you mean a bug in some other program that is using X and causes X to use unlimited memory?

If this is a bug in X then upgrading is probably easier than diagnosing it. If it is a bug in some program using X, I don't know whether upgrading would make a difference.

I don't know how to place limits on resources inside X. Hopefully the OP understood what you meant by that. That does seem to be the more reasonable thing to try first, in case this is a bug in something using X.