http://bugzilla.spamassassin.org/show_bug.cgi?id=3828





------- Additional Comments From [EMAIL PROTECTED]  2004-10-04 13:59 -------
quick note.  Alessandro uses -A flags on spamd, and -d flag on spamc just like i
do.  dunno if that plays any importance or not.

>> Can someone experiencing this issue post the following:

i think have provided most of what i can so far. 

stracing the spamd children are easy, but you will find there are no spamc
clients to strace because they have all went away due to the spamc timeout.  
dunno why spamd children still hang onto a socket with no client on the other
end??   my guess is since this is so random (2 hours to make 15 children stale
doing 10k/hour), the only time it is really occuring is when the spamc timeout
is encountered.   i can probably set the timeout to 1 sec and try that theory
myself.

>> ps auwwx (or -ef) output

see initial post.. note the start times.

>> strace the spamd pids

see initial post.. "strace the parent shows wait(), children show read()."

>> lsof the spamd pids

see comment #5 for lsof on 2 children.  you want full lsof or just filehandles?

>> strace the spamc pids

impossible or very hard to match spamc requests with which spamd server they
went to since it is round-robin.   also, i never see stale spamc clients because
of the spamc timeout.   guess i could up the timeout limit to something really
huge and try and catch it.

>> lsof the spamc pids

same thing as above.






------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

Reply via email to