#26 DGAV causing web browsing to suddenly die
Hi,
in logs, there is no exact evidence of what could cause this strange behaviour, even the possibility to set some process debug logging in dansguardian configuration.
Randomly after some time DGAV is running, whole web browsing in the LAN behaves like the service is unavailable. Here is an excerpt from /var/log/messages:
May 28 12:12:24 ns dansguardian: Error reading ip ipc. (Ignorable)
May 28 12:12:24 ns dansguardian: Can't read from unix socket: No such file or directory
May 28 12:12:26 ns dansguardian: Exception writing to url cache
May 28 12:12:26 ns dansguardian: Can't write to unix socket: Broken pipe
no strange entries in squid or in other log files found.
From my observation, DG does not handle it's processes correctly. The number of dansguardian processes keeps raising. Even if no web browsing in the LAN for a couple of hours, the number of DG processes stays around 70 ! Also in the arp cache, there are held entries of LAN coputers that were shutdown few hours ago.
After dansguardian restart (sometimes squid and clamav as well) all is back to normal, but only for a short period of time - like hours/days.
As far as i can remember, this behaviour started after upgrade of ClamAV from 0.88 to 0.90... And it has been observed on 2 different servers running Trustix Secure Linux 3.0.5 (kernel 2.6.19.7), DGAV 6.4.4.2, squid-2.5.STABLE14 and ClamAV 0.90.2
With this kind of behaviour is not acceptable for the regular usage of DGAV.
Should you have any questions, please, do not hesitate to let me know.
Thanks in advance
