Memory Usage in X (Fedora Core 2)

Chebyshev · 07-14-2004, 06:37 PM

Hello all:
I'm having some memory usage issues while running X and Fedora Core 2 with Gnome. The physical memory usage just goes up and up as the computer stays on. Restarting brings it back down to normal levels. I've looked at top and it says that a lot of RAM is used, but nothing seems to be using an exorbitant amount. Here it top output, sorted by memory usage:

top - 16:36:07 up 6:26, 2 users, load average: 1.08, 0.80, 0.55
Tasks: 81 total, 3 running, 78 sleeping, 0 stopped, 0 zombie
Cpu(s): 85.7% us, 14.3% sy, 0.0% ni, 0.0% id, 0.0% wa, 0.0% hi, 0.0% si
Mem: 1036664k total, 1026044k used, 10620k free, 125908k buffers
Swap: 2040244k total, 0k used, 2040244k free, 528116k cached

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
2702 root 16 0 117m 37m 88m R 63.2 3.7 31:42.21 X
14269 reede 15 0 79284 34m 22m S 0.0 3.4 15:05.80 python
15437 reede 15 0 97540 31m 30m S 0.0 3.1 0:34.46 firefox-bin
14328 reede 25 10 31588 18m 23m R 0.0 1.8 0:04.45 rhn-applet-gui
14260 reede 15 0 37088 17m 20m S 7.0 1.7 0:02.64 nautilus
14258 reede 16 0 21660 11m 18m S 0.0 1.2 0:01.98 gnome-panel
15515 reede 15 0 24640 11m 20m S 0.0 1.2 0:00.84 gnome-terminal
14165 reede 16 0 19908 9124 17m S 0.0 0.9 0:00.64 gnome-session
14322 reede 15 0 19532 9068 17m S 7.0 0.9 0:04.60 wnck-applet
14305 reede 15 0 20512 8620 18m S 0.0 0.8 0:00.69 mixer_applet2
14301 reede 16 0 19048 8180 17m S 0.0 0.8 0:00.56 clock-applet
14222 reede 16 0 20148 7808 18m S 0.0 0.8 0:00.69 gnome-settings-
14351 reede 15 0 39312 7716 11m S 0.0 0.7 0:20.62 xmms
14303 reede 16 0 18428 7124 17m S 0.0 0.7 0:00.27 notification-ar
14149 reede 16 0 10324 7056 4944 S 0.0 0.7 0:03.29 gconfd-2
14265 reede 16 0 18476 7008 17m S 0.0 0.7 0:00.44 eggcups
14220 reede 15 0 13356 6968 11m S 14.0 0.7 0:04.97 metacity
14267 reede 16 0 24412 6612 19m S 0.0 0.6 0:00.22 evolution-alarm
14256 reede 15 0 17888 6092 16m S 0.0 0.6 0:01.26 magicdev
2993 reede 16 0 25864 5384 20m S 0.0 0.5 0:00.09 evolution-womba
2316 ntp 16 0 4332 4332 3596 S 0.0 0.4 0:00.18 ntpd
14326 reede 16 0 12324 4216 11m S 0.0 0.4 0:00.23 pam-panel-icon
2388 xfs 16 0 5088 3544 2388 S 0.0 0.3 0:08.26 xfs
14274 reede 19 0 15388 3116 6776 S 0.0 0.3 0:00.03 gnome-vfs-daemo
2691 root 16 0 11296 3060 10m S 0.0 0.3 0:02.61 gdm-binary
2789 reede 16 0 6956 2828 6216 S 0.0 0.3 0:00.79 bonobo-activati
2335 root 16 0 7024 2788 6200 S 0.0 0.3 0:00.18 sendmail
7556 root 16 0 6820 2712 6216 S 0.0 0.3 0:00.21 bonobo-activati
2344 smmsp 16 0 6124 2364 5304 S 0.0 0.2 0:00.00 sendmail
14241 reede 15 0 4396 2352 3804 S 0.0 0.2 0:00.48 xscreensaver
2503 root 15 0 10536 2128 10m S 0.0 0.2 0:00.05 gdm-binary
2095 root 16 0 7612 2124 5936 S 0.0 0.2 0:00.39 cupsd
2284 root 18 0 3648 1452 3436 S 0.0 0.1 0:00.07 sshd
15517 reede 16 0 4400 1368 3948 S 0.0 0.1 0:00.03 bash
14227 reede 16 0 2508 1220 2308 S 0.0 0.1 0:00.06 fam
15432 reede 19 0 4340 972 3900 S 0.0 0.1 0:00.00 run-mozilla.sh
15428 reede 19 0 4080 960 3900 S 0.0 0.1 0:00.00 firefox
2299 root 15 0 2016 904 1684 S 0.0 0.1 0:00.00 xinetd
2426 dbus 17 0 1860 896 1700 S 0.0 0.1 0:00.00 dbus-daemon-1
15535 reede 15 0 1820 896 1620 R 14.0 0.1 0:00.06 top

You'll notice that I'm running gDesklets, an xterm and Firefox 0.9.1. Apart from that, I really only run Evolution once in a while for email. You'll also notice that almost all my physical RAM is used, while the swap file is completely empty. Anyone got any idea what would cause this? Thanks!

ilikejam · 07-15-2004, 06:54 PM

Hi.

There's most probably nothing wrong - Linux likes to use any available RAM for disk cache and buffers. This memory is instantly freed if it's needed.

Have a look at the output from 'free' in a console.
You should see something like this:

Code:

             total       used       free     shared    buffers     cached
Mem:        127212     123648       3564          0       1012      53392
-/+ buffers/cache:      69244      57968
Swap:       130872      10936     119936

The second number after +/- buffers/cache is the amount of RAM (in KB) you actually have for applications. You'll probably find you have several hundred MB left.

I'm running Fluxbox, XINE, Firefox and two xterms, and I've actually got about 57MB of RAM free, as you can see from the 57968 number above, but it looks like I've only got 3 MB left in 'top'.

Dave

foo_bar_foo · 07-15-2004, 07:31 PM

but this:
Cpu(s): 85.7% us, 14.3% sy, 0.0% ni, 0.0% id, 0.0% wa, 0.0% hi, 0.0% si
I have about the same amount of RAM as you and am right now running about the same app level
Me: (system up and running about 8 hours or so)
Tasks: 79 total, 1 running, 78 sleeping, 0 stopped, 0 zombie
and my cpu usage is (2.6 Ghz P4)
Cpu(s): 1.2% user, 0.8% system, 0.0% nice, 98.0% idle, 0.0% IO-wait

YOU HAVE
0.0 idle cpu

my mem usage:
Mem: 1033552k total, 269632k used, 763920k free, 19208k buffers
Swap: 2096472k total, 0k used, 2096472k free, 153648k cached

yours:
Mem: 1036664k total, 1026044k used, 10620k free, 125908k buffers
Swap: 2040244k total, 0k used, 2040244k free, 528116k cached

so yea you are running up alot of RAM as well but that's whatever is using up all your cpu cycles ??????
your X is using 63.2 % cpu cycles
my Xfree86 is using 3.7% cpu cycles (i'm running KDE)

MAN !! anybody have an explanation ??
or better yet other people with fairly new systems post results !

ilikejam · 07-15-2004, 08:12 PM

Here's the results from my server. It runs Folding@home constantly. It's also running KDE 3.2, Firefox, sshd, vsftpd and xawdecode.

Code:

top - 02:08:25 up 28 days, 18:41,  5 users,  load average: 1.18, 1.08, 1.02
Tasks:  69 total,   2 running,  67 sleeping,   0 stopped,   0 zombie
Cpu(s):  8.0% us,  0.7% sy, 90.4% ni,  0.0% id,  0.0% wa,  0.3% hi,  0.7% si
Mem:    256420k total,   242408k used,    14012k free,    41792k buffers
Swap:   265064k total,    24964k used,   240100k free,   110840k cached

(It's an Athlon 2500+)

Here's the output from 'free'

Code:

             total       used       free     shared    buffers     cached
Mem:        256420     243592      12828          0      41896     110992
-/+ buffers/cache:      90704     165716
Swap:       265064      24964     240100

As you can see, there's 165MB of Physical RAM left. and it's been running for nearly a month (KDE was restarted last week, and has been running ever since. The ftp and ssh servers have been on almost as long as the system's been running).

Dave

bcaka · 07-31-2004, 01:11 PM

I recently installed fedora and have found a similar situation - X consumes far more cpu cycles than I have ever seen before, usually more than 50-60% as soon as a user logs on. If I connect remotely with VNC, X goes to 100% cpu busy and stays there until I close the session.

Any ideas are appreciated, thanks.