On Mon, Oct 04, 2004 at 06:25:09PM -0300, Everton da Silva Marques wrote: > On Mon, Oct 04, 2004 at 04:51:08PM -0400, Christopher X. Candreva wrote: > > > > Just to update my own post: I've since found that clamd is shooting to 99% > > cpu quite often. I've identified a few files that cause this. > > > > For example, I've identified one particular 440k .exe file (aparently an > > SNES emulator) that takes almost 2 minutes to scan, pegging CPU usage at > > that time. As a comparison, the latest 12meg mozilla installer .exe scans > > in about 6 seconds. > > > > Production systems are Solaris 8 on sparc, but I've also confirmed on my > > linux desktop. > > > > Am I the only one seeing this ? > > I have a 6Mb e-mail file which takes 9-11 minutes to be > scanned with clamscan 0.80rc3 and latest cvs snapshots. > > Tested in Solaris 7 on Sparc.
I recently started getting mimedang sendmail Milter error state situations, which seemed to be a consequence of clamav taking several minutes or more to scan certain files. This is on multi-cpu 450MHz SPARC 420R systems running Solaris9. Increasing the milter-related timeouts from 1 minute to several minutes seemes to have eliminated many of the errors, however this is more fixing the symptom than the root cause of the problem. Certain files, often several MB though not necessarily, take minutes to scan, instead of seconds. On a off-topic side note, if anyone knows what SMTP related timeout issues come up if a Milter timeout is set to greater than several minutes, I'd be very interested to hear. Does sendmail somehow keep the SMTP session alive even if the Milter is taking longer than the SMTP DATA timeout might be, or am I restricted to the SMTP timeout periods? Mark -- Mark G. Thomas ([EMAIL PROTECTED]) http://www.misty.com/ http://mail-cleaner.com/ _______________________________________________ http://lists.clamav.net/cgi-bin/mailman/listinfo/clamav-users
