Re: lookup on memory shortage

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Fri, Oct 03, 2008 at 01:11:05PM +0200, Manuel Bouyer wrote:
> Hi,
> I got a lockup again. I had top running, here's what it displayed before
> the box wedged:
> 
> |load averages:  2.20,  1.24,  0.98;               up 3+14:52:30        
> 03:40:25
> |40 processes: 3 runnable, 35 sleeping, 1 zombie, 1 on CPU
> |CPU states:  0.0% user,  0.0% nice,  100% system,  0.0% interrupt,  0.0% idle
> |Memory: 294M Act, 144M Inact, 12M Wired, 11M Exec, 77M File, 16K Free
> |Swap: 256M Total, 256M Used, 4K Free
> |
> |  PID USERNAME PRI NICE   SIZE   RES STATE      TIME   WCPU    CPU COMMAND
> | 3744 root      43    0  5520K  348K RUN        2:50  1.10%  0.24% cc1
> |    1 root      85    0   748K    4K wait       0:49  0.00%  0.00% <init>
> | 5951 bouyer    43    0   756K  796K CPU        0:33  0.00%  0.00% top
> |  408 bouyer    85    0   764K  412K select     0:12  0.00%  0.00% 
> screen-4.0.3
> |  380 bouyer    85    0   824K  656K select     0:08  0.00%  0.00% sshd
> |  218 root      85    0  1780K 5672K pause      0:06  0.00%  0.00% ntpd
> |11203 root      85    0   748K    4K wait       0:05  0.00%  0.00% 
> <pbulk-build
> |  400 bouyer    43    0   756K   12K RUN        0:01  0.00%  0.00% 
> screen-4.0.3
> |  339 root      85    0   752K    4K kqueue     0:01  0.00%  0.00% <master>
> |24186 root      85    0   748K    4K kqueue     0:01  0.00%  0.00% <tail>
> |15925 root      85    0   124K  116K RUN        0:00  0.00%  0.00% sh
> |13806 root      85    0   752K  908K piperd     0:00  0.00%  0.00% cron
> |  364 root      85    0   756K  172K nanoslp    0:00  0.00%  0.00% getty
> |  326 root      85    0   756K  172K nanoslp    0:00  0.00%  0.00% getty
> |  395 root      85    0  1156K    4K pause      0:00  0.00%  0.00% <tcsh>
> |  375 bouyer    85    0  1084K    4K pause      0:00  0.00%  0.00% <tcsh>
> |  194 bouyer    85    0  1020K    4K pause      0:00  0.00%  0.00% <tcsh>
> 
> This time I don't understand where the memory has gone, because there's
> no big processes running (unless cc1 has grown a lot after the last top
> display, and before the box hanging).

Additionnal info: after reboot and looking at the pbulk log, it was
building the stalin, which is knows to produce a cc1 process which grows very
large. So the memory may have been in cc1, but top didn't have time to show
it.

-- 
Manuel Bouyer, LIP6, Universite Paris VI.           
Manuel.Bouyer%lip6.fr@localhost
     NetBSD: 26 ans d'experience feront toujours la difference
--