Hi, all!

Today I found one of my systems almost hung because of spamd 
processes that were using lots of CPU but apparently not doing
anything. This is on a Solaris box with Sendmail. Spamd is called via a
system-wide procmail script. After this happened twice I restarted
spamd and sendmail and turned on the -D debug flag.

Why would most messages get processed in a couple of seconds
while some take over 100? 

debug: running raw-body-text per-line regexp tests; score so far=1.9
debug: running full-text regexp tests; score so far=1.9
debug: Razor is available
debug: Razor Agents 1.17, protocol version 2.
debug: Read server list from /.razor.lst
debug: 255427 seconds before closest server discovery
debug: Closest server is 64.90.187.2
debug: Connecting to 64.90.187.2...
debug: Connection established
debug: Signature: 867705029d64c7bc0d600caf95ee6ac4e92fae88
debug: Server version: 1.11, protocol version 2
debug: Server response: Negative 867705029d64c7bc0d600caf95ee6ac4e92fae88
debug: Message 1 NOT found in the catalogue.
debug: Agent terminated
debug: trying Received header date for real time:       Mon, 1 Apr 2002 17:08:51 -0500
debug: no Received headers found, not raising flag
debug: is Net::DNS::Resolver unavailable? 0
debug: DNS MX records found: 0
debug: DNS MX records found: 0
debug: checking RBL orbs.dorkslayers.com., set relay
debug: checking RBL relays.osirusoft.com., set relay
debug: checking RBL relays.ordb.org., set relay
debug: checking RBL ipwhois.rfc-ignorant.org., set rfci
debug: checking RBL relays.visi.com., set relay
debug: checking RBL results in set relay for 127.0.0.6
debug: checking RBL results in set relay for 127.0.0.4
debug: AWL active, pre-score: 3.7, mean: undef
debug: Post AWL score: 3.7
debug: is spam? score=3.7 required=5.1
logmsg: clean message (3.7/5.1) for (unknown):60001 in 127 seconds.

I'd like to have the RBL checks turned on but not if they cause such
problems. Even so, it's not clear which one (if any) is causing the
problem. I've had these checks turned on for months, so it is not
clear to me that that is necessarily the problem. At the same time
as spamd goes nuts the lead sendmail process exits leaving all
children as orphans -- yet the syslog shows connections being
rejected because sendmail has exceeded its limit of incoming
connections. 

After the load average begins skyrocketing the spamassassin
debug log looks like this:

debug: running header regexp tests; score so far=0
logmsg: connection from localhost [ 127.0.0.1 ] at port 36430
debug: running body-text per-line regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
debug: running header regexp tests; score so far=0
logmsg: connection from localhost [ 127.0.0.1 ] at port 36431
debug: running body-text per-line regexp tests; score so far=0
debug: running header regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
logmsg: connection from localhost [ 127.0.0.1 ] at port 36474
debug: running header regexp tests; score so far=0
logmsg: connection from localhost [ 127.0.0.1 ] at port 36480
debug: running header regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
debug: running body-text per-line regexp tests; score so far=0
debug: running uri tests; score so far=1.9
debug: running uri tests; score so far=1.9

I'm now running last night's CVS release, by the way. Server
startup command was:

   /usr/local/bin/spamd -x -u nobody -D

Nothing odd appears in the /var/log/syslog or /var/adm/messages
files.

As a test I restarted with the "-L" parameter and the same thing
happened again, so I don't think the problem has to do with the
network checks. The stuck processes were racking up lots of
CPU time, but eventually seemed to finish their processing once
I stemmed the tide of incoming mail.

Anyone have a clue? I think it's back to 2.11 for me for now...

reb 


_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to