On Sat, 9 Mar 2002, Charlie Watts wrote: > On Fri, 8 Mar 2002, Duncan Findlay wrote: > > On Fri, Mar 08, 2002 at 02:18:25PM -0700, Charlie Watts wrote: > > > On Fri, 8 Mar 2002, Charlie Watts wrote: > > > > > > > spamd from last night sometimes doesn't always reap its children; This is > > > > with the -S setting on or off, with auto-whitelist on or off. > > > > > > > > They sit there sucking up CPU like there is no tomorrow. > > > > > > > > Went back to an older spamd, all is well. > > > > > > > > I'll try to troubleshoot this over the weekend. > > > > > > > > I'm using FreeBSD 4.5-stable. > > > > > > > > Can somebody think of another socket-serving perl daemon that spawns > > > > children and is reasonably portable? I'm just surprised that this > > > > continues to be a problem ... > > > > > > You know, I'm halfway wondering if this is a FreeBSD/Perl issue. > > > > > > I think I was wrong; I'm starting to see the spamd-doesn't-get-reaped > > > issue with my older spamd too. I did cvsup & make FreeBSD last night. > > > > Try the -m patch at bug 78 with -m set to something ridiculously high. The > > -m patch has a different SIGCHLD handler. > > Doesn't seem to help. I'm thinking it isn't a sighandling issue at all. <snip> > So I wonder if it is a non-terminating regex. Anybody else seeing anything > like this? > > If it is a regex, got any hints for how to determine which one it is?
Now I'm certain it's a slow regex, because I've found a message that takes a few minutes to process. I'm trying to narrow down the cause right now. It isn't a network test; It's still slow with -L. It's slow with "spamassassin" too. OK, it isn't a body test ... I've replaced the body with "token body" and it's still slow. Will report back when I narrow it down ... -- Charlie Watts [EMAIL PROTECTED] _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
