On 2016-05-13 09:43:24 [-0600], Will Aoki wrote: > After upgrading from 0.98.7+dfsg-0+deb7u1 to 0.99+dfsg-0+deb7u2 two > months ago, clamd on one of our servers has crashed approximately daily. > It's rarely stayed running for more than 24 hours. > > Before crashing, the daemon spews the message > > accept() failed: This is it? Nothing more? There should be an error message included after the colon.
> The kernel is not reporting segfaults or OOM. Correct. Based on the source code the `accapt()' function failed and therefore the daemon terminated. > On this particular server, clamd is used by clamav-milter. A Nagios > check script also runs clamdscan about every five minutes against a CAB, > an EXE, a bzip2'd EXE and a zip file that alll contain > "Clamav.Test.File-6". As of a Monday (long after the problem starte), > the script has started scanning another file we've had false-positive > problems with. You mean you gave a "normal" file and it was reported as a virus? Sebastian
