[EMAIL PROTECTED] wrote on 11/19/2004 08:09:26 PM:

> So, I have a new patch that should greatly improve memory usage
> if you have a low memory situation (where "low" == "< 100MB of
> free RAM", typically ;)
>
> It should apply cleanly to 3.0.0.   I'd be very interested in
> feedback from people trying it... here it is:
>
>   http://bugzilla.spamassassin.org/show_bug.cgi?id=3983
>
> It adds a couple of spamd command-line switches, but should do sane things
> if you use your existing switches as-is.
>
> --j.

The patch seems to be working pretty well on my system, don't notice any slowdowns and paging is definitely better. I'm running on a FreeBSD 5.2.1 system with 192MB of ram, using the defaults for the new switches, so two spamd child processes start up and it seems my system runs with just those two active most of the time.  I've seen it running with 5 processes a few times, which is what I have my limit set to, but most of the time it's using 2 or 3.  Here's a couple of top displays, the first is before the patch, the second is after.

last pid: 64134;  load averages:  1.02,  1.38,  1.30   up 45+20:26:52  09:34:27
28 processes:  1 running, 27 sleeping
CPU states: 74.3% user,  0.0% nice, 13.2% system,  0.4% interrupt, 12.1% idle
Mem: 116M Active, 9728K Inact, 47M Wired, 6664K Cache, 28M Buf, 572K Free
Swap: 512M Total, 48M Used, 464M Free, 9% Inuse, 316K In, 416K Out

  PID USERNAME   PRI NICE   SIZE    RES STATE    TIME   WCPU    CPU COMMAND
64073 spamd      113    0 53552K 38820K select   0:21 13.64% 12.40% perl
64011 spamd        4    0 56244K 37632K accept   0:51 15.61% 11.57% perl
63916 spamd      107    0 59760K 36248K select   1:28 15.15% 11.23% perl
63995 spamd        4    0 57212K 39360K accept   0:58 10.62%  9.91% perl
63996 spamd        4    0 56248K 37392K accept   0:46  4.13%  3.76% perl
  389 root        96    0  1608K   268K select   4:58  0.00%  0.00% ntpd
  260 root        96    0  1312K   248K select   3:47  0.00%  0.00% syslogd
  414 root        96    0  3512K   476K select   3:25  0.00%  0.00% sendmail
28604 spamd       20    0 46188K     0K pause    1:11  0.00%  0.00% <perl>
  431 root         8    0  1336K   120K nanslp   0:40  0.00%  0.00% cron
  359 root        96    0  1236K    84K select   0:18  0.00%  0.00% usbd
  408 root        96    0  3488K     0K select   0:10  0.00%  0.00% <sshd>
  417 smmsp       20    0  3404K     0K pause    0:08  0.00%  0.00% <sendmail>
64116 root        96    0  2212K  1076K RUN      0:00  0.00%  0.00% top
64094 root         4    0  6240K     0K sbwait   0:00  0.00%  0.00% <sshd>
64112 root        20    0  2312K     0K pause    0:00  0.00%  0.00% <csh>
sidewinder#


last pid: 74409;  load averages:  1.25,  0.85,  0.70   up 46+20:09:26  09:17:01
26 processes:  2 running, 24 sleeping
CPU states: 82.9% user,  0.0% nice, 14.8% system,  1.6% interrupt,  0.8% idle
Mem: 73M Active, 22M Inact, 47M Wired, 11M Cache, 28M Buf, 27M Free
Swap: 512M Total, 13M Used, 499M Free, 2% Inuse

  PID USERNAME   PRI NICE   SIZE    RES STATE    TIME   WCPU    CPU COMMAND
74322 spamd      127    0 56592K 46180K select   1:02 39.79% 39.79% perl
74370 spamd       -8    0 55752K 45212K piperd   0:38 35.92% 35.64% perl
  389 root        96    0  1608K   396K select   5:04  0.00%  0.00% ntpd
  260 root        96    0  1312K   280K select   3:52  0.00%  0.00% syslogd
  414 root        96    0  3512K   868K select   3:29  0.00%  0.00% sendmail
  431 root         8    0  1336K   192K nanslp   0:40  0.00%  0.00% cron
66431 spamd       96    0 46432K 28368K select   0:28  0.00%  0.00% perl
  359 root        96    0  1236K    84K select   0:18  0.00%  0.00% usbd
  408 root        97    0  3488K   488K select   0:10  0.00%  0.00% sshd
  417 smmsp       20    0  3404K   424K pause    0:09  0.00%  0.00% sendmail
66044 ajezierski  96    0  6240K  1052K select   0:01  0.00%  0.00% sshd
66047 root        20    0  2316K  1088K pause    0:00  0.00%  0.00% csh
66042 root         4    0  6240K   924K sbwait   0:00  0.00%  0.00% sshd
sidewinder#

Great Job!

Thanks
Andy

Reply via email to