0.75.1 was built on freebsd 5.2, gcc 3.3.3, and ran for months with never a hint of unhappiness. no crashes that i ever recall happening.
0.80 was built on freebsd 5.2, gcc 3.3.3, five days ago, and crashes many times a day.
i'm still having problems, unfortunately. if i allow too many concurrent incoming messages, clamd either freezes up or aborts after some amount of time, still. unfortunately, with limited concurrent incoming, messages accumulate on the MX gateways and eventually the trigger level is reached to have the queues dump directly to my pop server, bypassing all spam/virus scanning. so that's not a solution.
in desparation, last night i compiled gcc 3.4.3, and also installed binutils 2.15. recompiled clamav-0.80. this morning, after multiple pages advising me that the queue was going into bypass mode, i checked the server - the clamd process had grown (according to 'top') to 157 megs in size, approximately the same for resident.
this is really frustrating, particularly after coming off a trouble free X number of months of running .75. i'd gladly go back to .75 except for the small problem that that's not possible due to the changes that .80 addressed.
i dunno where to go from here.
here's a little more background info: server has 512M ram. dual PIII 900Mhz cpus. running under qmail-scanner. i've tried various softlimits, but they don't seem to be the problem. current softlimit is 60MB.
Paul Theodoropoulos http://www.anastrophe.com http://www.smileglobal.com
_______________________________________________ http://lists.clamav.net/cgi-bin/mailman/listinfo/clamav-users
