I noticed a problem for the last several days (2.4.13 Tyan PIII Dual)
but couldnt put my finger on it till i started paying attention:
kswapd is working furiously, but accomplishing nothing.
that is, kswapd is logging lots of CPU time, but no swap space is
being used. (see below).
i dont recall having problems until towards the end of the day (i
shutdown at nite). when i open or close a netscape session for
example, or do a dpkg blah-blah, the process hangs for a long time
before getting going. at one point prior to the 'top' below, i had
something like 500XXXK used, and still not a drop, err byte, of swap
used.
this behavior was not occuring with 2.4.12 (will check tmw) and
earlier versions.
les schaffer
================= INFO =================
speggy)~/: dmesg | grep swap
Starting kswapd
Adding Swap: 506036k swap-space (priority -1)
--------------
(speggy)~/: top
19:07:09 up 13:12, 3 users, load average: 1.68, 1.45, 1.24
69 processes: 66 sleeping, 3 running, 0 zombie, 0 stopped
CPU states: 14.4% user, 85.5% system, 0.0% nice, 0.1% idle
Mem: 513916K total, 443932K used, 69984K free, 7808K buffers
Swap: 506036K total, 0K used, 506036K free, 166992K cached
PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2873 godzilla 9 0 31984 30M 12272 S 0.0 6.0 0:00 mozilla-bin
[ mozilla-bin repeated ad infinitum)
356 godzilla 9 0 27528 26M 3688 S 0.0 5.3 4:16 xemacs
298 root 6 -10 202M 26M 4120 S < 0.9 5.1 9:04 XFree86
335 godzilla 9 0 7672 7668 6100 S 0.0 1.4 0:17 gabber
326 godzilla 9 0 4908 4908 3812 S 0.0 0.9 0:06 panel
[ etc ... ]
3 root 19 19 0 0 0 SWN 0.0 0.0 0:00 ksoftirqd_CPU0
4 root 19 19 0 0 0 SWN 0.0 0.0 0:00 ksoftirqd_CPU1
5 root 19 0 0 0 0 RW 95.9 0.0 282:40 kswapd
6 root 9 0 0 0 0 SW 0.0 0.0 0:00 bdflush
--------------
(speggy)~/: ps uxwa // about 50 minutes after 'top' above
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
root 5 37.9 0.0 0 0 ? SW 05:54 315:56 [kswapd]
--------------
(speggy)/proc/: cat swaps
Filename Type Size Used Priority
/dev/hda4 partition 506036 0 -1
I've checked 2.4.12 today, and it definitely does NOT have the same
problem as i reported for 2.4.13. so something went wrong in 2.4.13
...
from top:
18:11:57 up 9:38, 2 users, load average: 0.05, 0.03, 0.04
84 processes: 83 sleeping, 1 running, 0 zombie, 0 stopped
CPU states: 2.0% user, 0.7% system, 0.0% nice, 97.3% idle
Mem: 513912K total, 439012K used, 74900K free, 1580K buffers
Swap: 506036K total, 54320K used, 451716K free, 152492K cached
PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
365 godzilla 11 0 84052 60M 15904 S 0.2 12.1 27:15 mozilla-bin
...
294 root 12 -10 204M 28M 5120 S < 1.9 5.6 6:17 XFree86
352 godzilla 11 0 31708 28M 4304 S 0.4 5.6 5:34 xemacs
331 godzilla 9 0 3336 2460 2096 S 0.0 0.4 0:07 gabber
...
5 root 9 0 0 0 0 SW 0.0 0.0 0:01 kswapd
for comparison, from 2.4.13 yesterday:
19:07:09 up 13:12, 3 users, load average: 1.68, 1.45, 1.24
69 processes: 66 sleeping, 3 running, 0 zombie, 0 stopped
CPU states: 14.4% user, 85.5% system, 0.0% nice, 0.1% idle
Mem: 513916K total, 443932K used, 69984K free, 7808K buffers
Swap: 506036K total, 0K used, 506036K free, 166992K cached
PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
2873 godzilla 9 0 31984 30M 12272 S 0.0 6.0 0:00 mozilla-bin
...
356 godzilla 9 0 27528 26M 3688 S 0.0 5.3 4:16 xemacs
298 root 6 -10 202M 26M 4120 S < 0.9 5.1 9:04 XFree86
335 godzilla 9 0 7672 7668 6100 S 0.0 1.4 0:17 gabber
...
5 root 19 0 0 0 0 RW 95.9 0.0 282:40 kswapd
les schaffer
Please try the newest .14-pre.
Thanks
On Mon, 29 Oct 2001, Les Schaffer wrote:
> I noticed a problem for the last several days (2.4.13 Tyan PIII Dual)
> but couldnt put my finger on it till i started paying attention:
> kswapd is working furiously, but accomplishing nothing.
>
> that is, kswapd is logging lots of CPU time, but no swap space is
> being used. (see below).
>
> i dont recall having problems until towards the end of the day (i
> shutdown at nite). when i open or close a netscape session for
> example, or do a dpkg blah-blah, the process hangs for a long time
> before getting going. at one point prior to the 'top' below, i had
> something like 500XXXK used, and still not a drop, err byte, of swap
> used.
>
> this behavior was not occuring with 2.4.12 (will check tmw) and
> earlier versions.
>
> les schaffer
>
> ================= INFO =================
> speggy)~/: dmesg | grep swap
> Starting kswapd
> Adding Swap: 506036k swap-space (priority -1)
>
> --------------
>
> (speggy)~/: top
>
> 19:07:09 up 13:12, 3 users, load average: 1.68, 1.45, 1.24
> 69 processes: 66 sleeping, 3 running, 0 zombie, 0 stopped
> CPU states: 14.4% user, 85.5% system, 0.0% nice, 0.1% idle
> Mem: 513916K total, 443932K used, 69984K free, 7808K buffers
> Swap: 506036K total, 0K used, 506036K free, 166992K cached
>
> PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
> 2873 godzilla 9 0 31984 30M 12272 S 0.0 6.0 0:00 mozilla-bin
> [ mozilla-bin repeated ad infinitum)
> 356 godzilla 9 0 27528 26M 3688 S 0.0 5.3 4:16 xemacs
> 298 root 6 -10 202M 26M 4120 S < 0.9 5.1 9:04 XFree86
> 335 godzilla 9 0 7672 7668 6100 S 0.0 1.4 0:17 gabber
> 326 godzilla 9 0 4908 4908 3812 S 0.0 0.9 0:06 panel
> [ etc ... ]
> 3 root 19 19 0 0 0 SWN 0.0 0.0 0:00 ksoftirqd_CPU0
> 4 root 19 19 0 0 0 SWN 0.0 0.0 0:00 ksoftirqd_CPU1
> 5 root 19 0 0 0 0 RW 95.9 0.0 282:40 kswapd
> 6 root 9 0 0 0 0 SW 0.0 0.0 0:00 bdflush
>
> --------------
>
> (speggy)~/: ps uxwa // about 50 minutes after 'top' above
>
> USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
> root 5 37.9 0.0 0 0 ? SW 05:54 315:56 [kswapd]
>
> --------------
>
> (speggy)/proc/: cat swaps
> Filename Type Size Used Priority
> /dev/hda4 partition 506036 0 -1
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
> Please try the newest .14-pre.
will run pre6 all day today and then post results in the evening when
the problems usually happen.
however, i am pushing on it now, several Matlab (4) and Netscape (10)
and xemacs (17) sessions open, and kswapd seems very quiet and system
is very responsive, so things look very good.
les
08:50:24 up 16 min, 2 users, load average: 0.44, 0.37, 0.23
179 processes: 178 sleeping, 1 running, 0 zombie, 0 stopped
CPU states: 1.2% user, 6.2% system, 0.0% nice, 92.6% idle
Mem: 513920K total, 510980K used, 2940K free, 2892K buffers
Swap: 506036K total, 4140K used, 501896K free, 93564K cached
PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
300 root 6 -10 232M 56M 6164 S < 1.3 11.1 0:21 XFree86
10036 godzilla 13 5 44668 42M 9388 S N 0.0 8.4 0:04 matlab
10037 godzilla 12 5 44668 42M 9388 S N 0.0 8.4 0:00 matlab
10038 godzilla 15 9 44668 42M 9388 S N 0.0 8.4 0:00 matlab
10039 godzilla 15 10 44668 42M 9388 S N 0.0 8.4 0:00 matlab
10040 godzilla 13 5 44668 42M 9388 S N 0.0 8.4 0:00 matlab
10041 godzilla 13 5 44668 42M 9388 S N 0.0 8.4 0:01 matlab
10042 godzilla 13 5 44668 42M 9388 S N 0.0 8.4 0:00 matlab
10043 godzilla 13 5 44668 42M 9388 S N 0.0 8.4 0:00 matlab
10045 godzilla 13 5 44668 42M 9388 S N 0.0 8.4 0:00 matlab
10046 godzilla 13 5 44668 42M 9388 S N 0.0 8.4 0:00 matlab
5 root 9 0 0 0 0 SW 0.0 0.0 0:00.13 kswapd