Hi, I found that even on systems here handling very light volume, after several hours clamd (0.91.1) goes to 200%-300% CPU (4 CPU system) and system load to 9.0 or more. I reverted to 0.90.3 and the problem went away.
When clamd got in this state, it was still scanning, just slowly, and I had to "kill -9" the process to get it to die. On Tue, Jul 17, 2007 at 09:42:05AM -0700, Tom Bombadil wrote: > Is anybody else experience a much higher load with 0.91 compared to 0.90.3? > > I'm running clamd over TCP. The config files and build options are > exactly the same: > > ./configure --sysconfdir=/opt/etc \ > --localstatedir=/var \ > --enable-bigstack \ > --enable-readdir_r \ > --with-dbdir=/var/clamav/db \ > --prefix=/opt/$CLAMAV_VERSION I'm running Solaris-9 on a handful of Sun 420R 4CPU 4GB RAM systems, with mimedefang and sendmail. ./configure --sysconfdir=/var/sendmail/etc --with-dbdir=/var/sendmail/clamav-db --disable-clamav --with-user=mail --with-group=mail --prefix=/opt/clamav-0.91.1 % egrep -v '#|^$' clamd.conf LogSyslog yes LogFacility LOG_LOCAL6 LogVerbose yes PidFile /var/run/clamd.pid DatabaseDirectory /var/sendmail/clamav-db LocalSocket /var/spool/MIMEDefang/clamd.sock FixStaleSocket yes MaxConnectionQueueLength 30 StreamMaxLength 20M MaxThreads 10 User mail AllowSupplementaryGroups yes ArchiveMaxFileSize 10M I take it the thing to do now is: Recompile with "-g" and try attaching with gdb to clamd when this happens, and get a backtrace of _all_ threads. Mark -- Mark G. Thomas ([EMAIL PROTECTED]) voice: 215-591-3695 http://mail-cleaner.com/ _______________________________________________ Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net http://lurker.clamav.net/list/clamav-users.html
