Tilman Schmidt wrote: > I have ClamAV running on several Linux mailservers. All of them stopped > working last night with similar symptoms: > > - Some time after 23h CEST (21h GMT) freshclam started complaining it > couldn't connect any update server. > > Apr 6 23:07:06 lx1 freshclam[15939]: nonblock_connect: connect timing > out (30 secs) > Apr 6 23:07:06 lx1 freshclam[15939]: Can't connect to port 80 of host > db.de.clamav.net (IP: 62.26.160.3) > Apr 6 23:07:06 lx1 freshclam[15939]: Trying host db.de.clamav.net > (62.201.161.84)... > > [repeating every 30 seconds with varying IP addresses] > This in itself isn't normally a reason for concern. > > - Five minutes later it gives up on incrementals and switches to > main.cvd, which is probably standard behaviour, but the connection > problems persist: > > Apr 6 23:12:08 lx1 freshclam[15939]: Incremental update failed, > trying to download main.cvd > Apr 6 23:12:38 lx1 freshclam[15939]: nonblock_connect: connect timing > out (30 secs) > Apr 6 23:12:38 lx1 freshclam[15939]: Can't connect to port 80 of host > db.de.clamav.net (IP: 195.246.234.199) > Apr 6 23:12:38 lx1 freshclam[15939]: Trying host db.de.clamav.net > (212.1.60.18)... > Apr 6 23:13:08 lx1 freshclam[15939]: nonblock_connect: connect timing > out (30 secs) > > - Some time later ClamAV complains it cannot update its database, and > exits: > > Apr 6 23:15:28 lx1 clamav-milter[15949]: Unable to lock database > directory > Apr 6 23:15:28 lx1 clamav-milter[15949]: Failed to load updated database > Apr 6 23:15:31 lx1 clamav-milter[15947]: ClamAv: mi_stop=1 > Apr 6 23:15:31 lx1 clamav-milter[15947]: Stopping ClamAV > 0.92.1/6635/Sun Apr 6 18:29:31 2008 > > Or on a different machine using MIMEdefang instead of clamav-milter: > > Apr 6 23:49:10 monolith clamd[4648]: reload db failed: Unable to lock > database directory (try 3)
It looks like freshclam locked the directory. This will no longer be a problem with 0.93, since no locking of the directory is required anymore! > Apr 6 23:49:10 monolith clamd[4648]: reload db failed: Unable to lock > database directory > Apr 6 23:49:10 monolith clamd[4648]: Terminating because of a fatal > error. > Simple question: why did that happen? IMHO a failure to update the > signatures, even if it persists for several hours, should not prevent > the continued use of the scan service with the signatures it already > has. Is this: > - a misconfiguration (ie. my own fault)? Run clamd under a wrapper that monitors it, and restarts it in case of failure. > - a bug? Failure to reload the DB shouldn't be a fatal error, please open a bug. > - a feature? No. Best regards, --Edwin _______________________________________________ Help us build a comprehensive ClamAV guide: visit http://wiki.clamav.net http://lurker.clamav.net/list/clamav-users.html
