gluck is too loaded
Hi,
gluck is suffering from too high load. Constantly, the load is about 4
now. One top snapshot shows (with the first indicator pending between
3 and 5):
03:47:15 up 62 days, 18:51, 7 users, load average: 3.58, 4.17, 4.34
185 processes: 180 sleeping, 4 running, 0 zombie, 1 stopped
CPU states: 36.3% user, 22.2% system, 0.0% nice, 41.5% idle
Mem: 3879936K total, 3867316K used, 12620K free, 128756K buffers
Swap: 2000084K total, 191444K used, 1808640K free, 2693140K cached
PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
3850 manty 19 0 112M 112M 1072 R 88.5 2.9 5:16 jigdo-file
13668 joy 16 0 242M 131M 131M R 18.7 3.4 65:05 lintian
31648 djpig 10 0 9140 9140 1452 S 16.9 0.2 1:31 urlcheck.py
31651 djpig 10 0 3316 3316 1452 S 3.7 0.0 0:39 urlcheck.py
20593 www-data 9 0 2476 2132 1932 S 2.7 0.0 0:00 apache
31653 djpig 9 0 3620 3620 1448 S 2.5 0.0 0:16 urlcheck.py
20778 www-data 9 0 2624 2624 1328 S 2.3 0.0 0:00 viewcvs.cgi
19272 www-data 9 0 2540 2232 2028 S 1.5 0.0 0:00 apache
16820 aba 12 0 1352 1352 1048 R 1.1 0.0 0:01 top
19712 www-data 9 0 2532 2284 2040 S 1.1 0.0 0:00 apache
19741 www-data 9 0 2584 2332 2096 S 0.9 0.0 0:00 apache
31650 djpig 9 0 3524 3524 1452 S 0.7 0.0 0:17 urlcheck.py
19693 www-data 8 0 2516 2220 1988 S 0.7 0.0 0:00 apache
19739 www-data 9 0 2524 2276 2024 S 0.7 0.0 0:00 apache
It seems to me that these processes are io-bound.
I think we need to not have a constant load of more than 2 on gluck.
Otherwise, it's a bit hard to use it as "general shell machine". So,
my question is: Can some of these long running cronjobs moved to other
machines and/or niced down? Also, would it be possible to allow the
usage of user directories for web access on merkel? Then some more
things could be done there instead of gluck (and I could get rid of
one rsync cron job ;).
Cheers,
Andi
--
http://home.arcor.de/andreas-barth/
PGP 1024/89FB5CE5 DC F1 85 6D A6 45 9C 0F 3B BE F1 D0 C5 D1 D9 0C
Reply to: