On Wed, 2005-11-16 at 21:55 +0000, Stephen Gran wrote: > On Wed, Nov 16, 2005 at 04:45:51PM -0500, Dale Blount said: > > On Wed, 2005-11-16 at 18:39 +0000, Stephen Gran wrote: > > > Would it be possible to attach gdb or strace to one of these hung > > > freshclams before killing them in the future? I have an idea that the > > > problem may be that attempts to download updates that timeout aren't > > > getting properly terminated, leaving freshclam in a confused state, but > > > I am currently unable to reproduce this myself, so am looking for people > > > experiencing the problem to help. > > > > I attached a strace to one of them but it didn't seem to do anything... > > I will try gdb next time. > > Even knowing that it's stuck in a read instead of a sleep would be > useful, for instance, but gdb will likely get us more info anyway.
It (finally?) happened again, here's the debugging data I've collected so far. I've left it running incase you need me to attach gdb to it (please provide gdb commands you'd like me to run). ------------------ This is from freshclam.log (notice the missing line about daily.cvd) Received signal: wake up ClamAV update process started at Wed Jan 11 06:02:00 2006 main.cvd is up to date (version: 35, sigs: 41649, f-level: 6, builder: tkojm) There's a file in /var/lib/clamav that seems to be a partial db download. # strings clamav-126c8db470982f9e | head -1 ClamAV-VDB:11 Jan 2006 11-19 +0100:1238:471:7:fc0540abba660046d63c07dfc921c0b9:IoGay674Et3R9y4zz94lJnSl8Ne23OpSvKZ6nu6yVL7ApfWPTOHeyPCHfd3QUnySG2Z8ffspKKepOEhzeOfK4uPvlGwe7K9I4XXS7+ly8Uu8kZdim+Bfqn2SvrODdE2RDlnau79HW7+UrM+6t7kCmV4+lUMYdy47OgYQ3yKyvWg:diego:1136974746 # strings daily.cvd | head -1 ClamAV-VDB:10 Jan 2006 16-53 +0100:1237:470:7:044a4eb33c5e260eb9f53cae2f33d6e0:gH5mMRmpRBFB1dKUTWDSKHfFw4F3HKu3VwDvZF6u0Uk0VzUUkhH6R+K1KVjoNihdhpsGGGXH1VAivAcxMtLuC0CghYIIzqJ0npJKlk2pRyaZhYiYBrH8LdN1KmFbHJzpybaJkLSyRHpDLgA96kxnKNkS/fa/Y0L4i+QzFWClF4b:diego:1136908400 This file matches closely to the date/time that freshclam started. It's 3 minutes off and only 1219 bytes, maybe the mirror it chose was having problems? -rw-r--r-- 1 clamav clamav 1219 2006-01-11 06:05 clamav-126c8db470982f9e There's still a connection to the update server. If I browse to mojave.daemon.be it loads quickly now, but I do not know if it did at 6am. # netstat -ap | grep freshclam tcp 0 0 myhost:38721 mojave.daemon.be:www ESTABLISHED 17435/freshclam Here's the output from strace as requested way back when. I left it connected for quite a while and it didn't move. # ps axuw | grep freshclam clamav 17435 0.0 0.1 2304 1044 ? Ss Jan10 0:00 /usr/bin/freshclam -p /var/run/clamav/freshclam.pid -d # strace -ff -p 17435 Process 17435 attached - interrupt to quit read(1, More debug information: # ps axl | grep freshclam 1 64 17435 1 16 0 2304 1044 348740 Ss ? 0:00 /usr/bin/freshclam -p /var/run/clamav/freshclam.pid -d # cat /proc/17435/status Name: freshclam State: S (sleeping) SleepAVG: 88% Tgid: 17435 Pid: 17435 PPid: 1 TracerPid: 10492 Uid: 64 64 64 64 Gid: 64 64 64 64 FDSize: 32 Groups: 64 VmSize: 2304 kB VmLck: 0 kB VmRSS: 1044 kB VmData: 200 kB VmStk: 16 kB VmExe: 36 kB VmLib: 1972 kB VmPTE: 16 kB Threads: 1 SigPnd: 0000000000000000 ShdPnd: 0000000000000000 SigBlk: 0000000000000000 SigIgn: 0000000000000000 SigCgt: 0000000180014003 CapInh: 0000000000000000 CapPrm: 0000000000000000 CapEff: 0000000000000000 Thanks, Dale _______________________________________________ http://lurker.clamav.net/list/clamav-users.html
