-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 David Goldsmith wrote:
> Check your mail log for error messages like this one: > > spamd[12960]: prefork: server reached --max-children setting, consider > raising it > > We've been running spamd with '-m8' (max children spawned) for quite > sometime and all of a sudden yesterday, we started getting similar > behavior where email was coming through without SA headers. > > I'm guessing that some of the network checks we are doing are taking > longer thus tying up the spawned spamd child processes longer. I bumped > our -m arg from 8 to 12 (still got the error) and then to 24 -- that > seems to have helped. Ok, I've dug into this some more because we've suddenly been having a lot of problems with this. Searching for references to that error message, I came across this old post -- http://www.nabble.com/Spamd-child-states--t2223988.html I grepped through our maillog for 'child states' and saw this: Feb 23 15:04:44 iceman14 spamd[12960]: prefork: child states: BBBBBBBBBBBBBBBBBBBBBBBB Feb 23 15:04:47 iceman14 spamd[12960]: prefork: child states: BBBBBBBBBBBBBBBBBBBBBBBB Feb 23 15:04:53 iceman14 spamd[12960]: prefork: child states: BBBBBBBBBBBBBBBBBBBBBBBB Feb 23 15:05:07 iceman14 spamd[12960]: prefork: child states: BBBBBBBBBBBBBBBBBBBBBBBB Feb 23 15:05:22 iceman14 spamd[12960]: prefork: child states: BBBBBBBBBBBBBBBBBBBBBBBB Feb 23 15:05:28 iceman14 spamd[12960]: prefork: child states: BBBBBBBBBBBBBBBBBBBBBBBB Feb 23 15:05:35 iceman14 spamd[12960]: prefork: child states: BBBBBBBBBBBBBBBBBBBBBBBB Feb 23 15:05:44 iceman14 spamd[12960]: prefork: child states: BBBBBBBBBBBBBBBBBBBBBBBB Feb 23 15:05:49 iceman14 spamd[12960]: prefork: child states: BBBBBBBBBBBBBBBBBBBBBBBB Feb 23 15:05:59 iceman14 spamd[12960]: prefork: child states: BBBBBBBBBBBBBBBBBBBBBBBB Feb 23 15:06:02 iceman14 spamd[12960]: prefork: child states: BBBBBBBBBBBBBBBBBBBBBBBB Feb 23 15:06:03 iceman14 spamd[12960]: prefork: child states: BBBBBBBBBBBBBBBBBBBBBBBB Doesn't look good. I went looking for other errors in the log and saw timeout errors involving ixhash: spamd[29382]: ixhash timeout reached at /etc/mail/spamassassin/iXhash.pm line 91. Is anyone experiencing problems connecting to the iXHash servers? I removed the iXhash.cf and iXhash.pm files from /etc/mail/spamassassin and restarted it. Now our child state log entries look like: Feb 23 15:32:09 iceman14 spamd[29656]: prefork: child states: BI Feb 23 15:32:10 iceman14 spamd[29656]: prefork: child states: IB Feb 23 15:32:14 iceman14 spamd[29656]: prefork: child states: II Feb 23 15:32:23 iceman14 spamd[29656]: prefork: child states: BB Feb 23 15:32:23 iceman14 spamd[29656]: prefork: child states: BBB Feb 23 15:32:23 iceman14 spamd[29656]: prefork: child states: BBBB Feb 23 15:32:23 iceman14 spamd[29656]: prefork: child states: BBBBB Feb 23 15:32:23 iceman14 spamd[29656]: prefork: child states: BBBBBB Feb 23 15:32:23 iceman14 spamd[29656]: prefork: child states: BBBBBBB Feb 23 15:32:23 iceman14 spamd[29656]: prefork: child states: BBBBBBBI Feb 23 15:32:25 iceman14 spamd[29656]: prefork: child states: IBBBBBBI Feb 23 15:32:28 iceman14 spamd[29656]: prefork: child states: IBBBBIBI Feb 23 15:32:28 iceman14 spamd[29656]: prefork: child states: IBBBBIBK Feb 23 15:32:28 iceman14 spamd[29656]: prefork: child states: IBIBBIB Feb 23 15:32:28 iceman14 spamd[29656]: prefork: child states: IBIBBB Feb 23 15:32:28 iceman14 spamd[29656]: prefork: child states: IIIBBB Feb 23 15:32:28 iceman14 spamd[29656]: prefork: child states: IIBBB Feb 23 15:32:28 iceman14 spamd[29656]: prefork: child states: IIBIB Feb 23 15:32:28 iceman14 spamd[29656]: prefork: child states: IIBKB Feb 23 15:32:29 iceman14 spamd[29656]: prefork: child states: IIBI Feb 23 15:32:29 iceman14 spamd[29656]: prefork: child states: IIBK Feb 23 15:32:31 iceman14 spamd[29656]: prefork: child states: III Feb 23 15:32:31 iceman14 spamd[29656]: prefork: child states: II Feb 23 15:32:37 iceman14 spamd[29656]: prefork: child states: BI Feb 23 15:32:38 iceman14 spamd[29656]: prefork: child states: II Periodic spikes as bursts of messages come through but then the children spamd processes get cleaned up. David Goldsmith -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3rc2 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFF3w/h417vU8/9QfkRArs3AKC0E9aNpuYVrjOycoRY6lf1U7lzFACeLgfD YCb+3YEV9iBun2PNgjdVeOA= =oEu1 -----END PGP SIGNATURE-----
