[Gllug] Memory usage when idle

Nix nix at esperi.org.uk
Sat May 8 13:39:26 UTC 2004


On Wed, 5 May 2004, Daniel P. Berrange spake:
> Have you disabled the 10's of background processes most distros start
> automatically like cron, gpm, atd, lpd, etc, etc, etc. These can
> have a surprisingly big impact.

The evidence is against you.

The daemons you mentioned:

 RSS   VSZ COMMAND
 400  2044 lpd Waiting  
 476  1500 /usr/sbin/crond
 432  1432 /usr/sbin/atd -l 2.4
 280  1592 gpm -m /dev/psaux -t ps2 -3 -R

God, 1.5Mb paged in (on a heavily loaded 128Mb box). I can see that
*that*'s going to break the bank. (But stopping those processes *will*
break some fairly important things, from locate(1) to Usenet news
processing. Stopping cron is particularly silly.)

Here's a bigger pile of daemons on that heavily loaded box (picked by
looking for processes with no controlling terminal, and then trimming
away things like random X apps run by users):

  RSS   VSZ COMMAND
  332  1400 init [3]       
  684  1832 /usr/sbin/syslog-ng -p /var/run/syslog-ng.pid
  452  1432 /usr/sbin/inetd
 2040  3908 /usr/sbin/named -u daemon
  224  1764 /usr/bin/esd -public -tcp -nobeeps -port 2574 -as 20 -promiscuous
  372  2616 /usr/bin/zsh /usr/sbin/mp3d
  372  2616 /usr/bin/zsh /usr/sbin/mp3d
  556  2936 /usr/sbin/sshd
  404  2044 lpd Waiting  
  364  1424 /usr/sbin/rpc.portmap
  744  1688 /usr/sbin/rpc.mountd
  688  1648 /usr/sbin/rpc.statd
  360  1428 /usr/sbin/rpc.rwalld
  364  1536 /usr/sbin/rpc.rquotad
 1148  3552 /usr/lib/news/bin/innd -p4 -i0 -c28
  236  1564 /usr/lib/news/bin/overchan
  476  1500 /usr/sbin/crond
  432  1432 /usr/sbin/atd -l 2.4
  388  1396 /usr/sbin/uptimed
  284  1592 gpm -m /dev/psaux -t ps2 -3 -R
  552  8920 /usr/bin/postmaster -B 256 -D/home/postgres/data -i -o -e
  272  1388 logger -t postgres -p local0.notice
  456 13908 /usr/libexec/mysqld -u postgres
  372  7556 /usr/X11R6/bin/xfs
  448  9912 postgres: stats buffer process                           
  448  8920 postgres: stats collector process                        
  456 13908 /usr/libexec/mysqld -u postgres
  456 13908 /usr/libexec/mysqld -u postgres
  180  5332 /usr/bin/perl /usr/bin/speechd -s festival -H hades
  444  2368 /bin/sh /etc/rc.d/rc.M
  560  2368 /bin/sh /etc/rc.d/rc.M
  860  5508 sshd: compiler at pts/1
  264  2580 ssh-agent -s
  764  5504 sshd: nix at notty
  712  1804 -amaterasu.srvr.nix LIST                                                  
 5148 43220 /usr/bin/snort -i adsl-phys -l /var/log.real/snort/adsl -k all -d -u snort -g snort -c /etc/snort/snort.conf
  304  1388 logger -p daemon.info -t snort
 7428 45636 /usr/bin/snort -i gord-phys -l /var/log.real/snort/gordianet -k none -d -u snort -g snort -c /etc/snort/snort.conf.internal
  336  1388 logger -p daemon.info -t snort
 1364  5504 sshd: root at pts/2

I think it would be mendacious to describe any of these as particularly
bloated: mysql and the pair of snorts are by far the worst offenders,
and the various sshds and PostgreSQL are a bit hefty. But add all that
up and it comes to a resident memory usage of 32Mb --- and that's
assuming that all pages are unshared, which is definitely untrue for
e.g. the mysqlds and sshds: the actual value is a few Mb lower.

On most boxes, that'll vanish in the noise. Even on this machine, it's
irrelevant enough that I've got two user-mode-linux partitions and a
large compilation running on there at the same time, and I'm not
swapping much (or at all, most of the time).

-- 
`If you believe in strong AI, then death is no longer a mystery,
 but merely a lack of adequate backups.' --- Steven McDougall
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug




More information about the GLLUG mailing list