Stephen Gran <[EMAIL PROTECTED]> writes: > Right. As above - there likely was no problem with clamd, but with > milter -> clamd communication.
I suspect that as well. > Unfortunately, we only have sendmail error messages, so it makes it > difficult to be sure where the problem was. If you like, put > LogSyslog in clamd.conf - that will let the milter log there as > well. I've done this now, thanks. > It looks the milter got stupid, and couldn't figure out how to talk to > clamd. Once you restarted it, it works normally again (and clamd is > busily creating new threads to handle the requests, as it should). I'll try restarting only clamav-milter next time, to rule out any problem with clamd. > If it continues to act up on you, put LogVerbose and LogClean in > clamd.conf - that will let you see the milter error messages, and also > clamd will log connections, so you can see where it's failing. If you > do this, beware that the milter log code is horrendous, and you will > want to log to syslog (with LogSyslog, as above) - letting it log > directly to the clamav.log will produce wierd results and possibly lost > data. > > This looks like it is not a soft limit problem, but a wierd state for > the milter. It would have been helpful if you had strace'd the milter > as well as clamd, but if it happened once, it may happen again, so you > may yet get the chance :( Yes, I see it often, so I suspect it won't go away this easily. I added LogVerbose and LogClean now. Thanks, Simon -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
