On 2009-03-04 17:49, Jorge Valdes wrote: > Because I run clamd via daemontools, whenever clamd crashes, its brought > back to life within a few seconds and things hum along fine until the > next crash, which is in my case, very unpredictable since I have gone > days without a crash, and suddenly have 3-5 crashes per day. This > production server is running 0.94.2 and scans 70000-80000 msgs/day. >
Hi, Here is the bugreport tracking this issue: https://wwws.clamav.net/bugzilla/show_bug.cgi?id=1370 So far we've been unable to reproduce the crash ourselves. As suggested in that bugreport, there are 2 things you can try to help us debug this issue: 1) Run clamd under valgrind and watch for crashes (valgrind --trace-children=yes --log-file=valgrind.log /usr/sbin/clamd) 2) Compile ClamAV with mudflap: ./configure CFLAGS="-O2 -g -D_FORTIFY_SOURCE=2 -fstack-protector -fmudflapth" LDFLAGS="-lmudflapth -pthread" Then run clamd in Foreground mode, and redirect output to a logfile: /usr/sbin/clamd >clamdlog 2>&1 However both valgrind and mudflap will make clamd run MUCH slower (10-20x), so I don't recommend doing this on a production server. Perhaps you can set up a valgrind/mudflap clamd on another box, and send mails to both the "normal" clamd, and the one running under valgrind/mudflap. Best regards, --Edwin _______________________________________________ Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net http://www.clamav.net/support/ml
