I believe I have the problem in control, thanks to help from this list.
jyoti_at_qcav01.enet.dec.com (an old Clemson Student - we do a good job!)
and nicolis_at_celfi.phys.univ-tours.fr are the ones who probably hit the
nail on the head with their comments about xdm configuration. The xdm
man page, and our recent explosion of X use as our students have
discovered roll-you-own-dial-up-IP-access tricks like tia and term/linux
and slipknot has me convinced that this is was the likely fix.
However - a bit before the xdm suggestions started rolling in, Dave Cherkus
suggested that if someone was using /proc, they might be causing the
problem. I umounted /proc, and nobody screamed, so if anyone was using
it, it was probably a hacker waiting for me to type in the root password :-).
At any rate, those are the two things that I did, and my seat of the pants
feeling is that the problem went away. System degredation has stopped,
and after I reboot I hope to see that the Kernel (which was burgeoning
rapidly) will not grow crazily.
Thanks everyone!
SUMMARY FOLLOWS:
From: "John P. Speno" <speno_at_cc.swarthmore.edu>
> It runs slap out of swap space.
> On reboot, the kernel's RSS is around 608k, and VSZ is about 1.77m.
> Here I am, one day after reboot, and RSS = 29m, and VSZ = 317m.
>
> Yesterday, before I rebooted, while the system was unusable, RSS was 85m
> and VSZ was 324m.
Yow! How about an upgrade to 3.2? I guess this doesn't have anything to do
with /sbin/swapdefault. That's all from me. Luck!
From: gachamb_at_milp.jsc.nasa.gov
there is a know problem with 3.2 kernel leak using a fddi card. are yo using
one?? if so contact dec and they will provide a patch.
From: sweber_at_mr.picker.com (Stephen Weber)
If you get response on this issue would you
please forward to me? We have seen the same problem
with version 1.3 (although a few days to run out
of swap space). I was hoping 3.0 would be the
answer to all our problems. ;) A side note,
we had a system with a typo in /etc/fstab which
did not allocate swap space properly, but that
showed in the swapon -s command.
From: Dave Cherkus <cherkus_at_UniMaster.COM>
System Janitor writes:
|> On reboot, the kernel's RSS is around 608k, and VSZ is about 1.77m.
|> Here I am, one day after reboot, and RSS = 29m, and VSZ = 317m.
|>
|> Yesterday, before I rebooted, while the system was unusable, RSS was 85m
|> and VSZ was 324m.
That's pretty darn big.
Did you do 'vmstat -M'? It might show something interesting.
The only leak I heard of was related to having a program
that continously scanned the /proc file system (i.e. some
sort of monitor). I forget which program it was that did
this, but it was not a Digital supplied program. There is
a patch for this - see
http://www.service.digital.com and
use the patch scan engine.
From: nicolis_at_celfi.phys.univ-tours.fr
Yes it's the infamous swap space problem. Add the following line to
the /usr/lib/X11/xdm/xdm-config file:
DisplayManager._0.terminateServer: true
This command tells the system to liberate used space.
From: anthony baxter <anthony.baxter_at_aaii.oz.au>
Yeah, there's a bug. Check ftp.service.digital.com for the patches readmes,
then hassle DEC to get the patch.
From: Jyotinath 17-Apr-1995 1015 +0800 <jyoti_at_qcav01.enet.dec.com>
Mike,
A couple of things might help in reducing memory consumption and
getting better performance:
1. cd /sbin
mv swapdefault swapdefault.orig
reboot (takes effect on reboot)
2. cd /usr/lib/X11/xdm
vi Xservers
===> change transport to "local:0" and add options
"-su -bs -terminate" to the last line
restart Xserver to take effect
From: judith_at_npac.syr.edu
Did you get a response to this problem? Is it the classic swapdefault file
problem, or do you already have it set for lazy swap?
We have a group of osf 3.0 alphas which periodically get into
"swap space below 10% free" and hang up, even though I am using lazy swap -
I'm wondering if anyone else is seeing this...
Received on Tue Apr 18 1995 - 14:56:17 NZST