We use the snapshot creation toolkit from rse@FreeBSD.org http://people.freebsd.org/~rse/snapshot/ on our Sophos PureMessage Server Farm (six FSC rx300 server). Cause PMX only supports FreeBSD-6.1, we can't upgrade to 6.2 at the moment. During creation of an snapshot the system panics from time to time Perhaps this is correlated with PR kern/68757. 115573 : cat /var/crash/info.13 Dump header from device /dev/da0s1b Architecture: i386 Architecture Version: 2 Dump Length: 1073242112B (1023 MB) Blocksize: 512 Dumptime: Fri Aug 17 13:01:36 2007 Hostname: rx300.hrz.uni-bielefeld.de Magic: FreeBSD Kernel Dump Version String: FreeBSD 6.1-RELEASE-p12 #0: Tue Jan 16 16:58:07 CET 2007 root@rx300.hrz.uni-bielefeld.de:/opt/tmp/obj/usr/src/sys/PMX Panic String: lockmgr: thread 0xc4c00d80, not exclusive lock holder 0xc4dd7c00 unlocking Dump Parity: 1320559640 Bounds: 13 Dump Status: good 115570 : df -kl Filesystem 1K-blocks Used Avail Capacity Mounted on /dev/da0s1a 507630 185360 281660 40% / devfs 1 1 0 100% /dev procfs 4 4 0 100% /proc /dev/da0s1d 16244334 2434562 12510226 16% /var /dev/da0s1e 4058062 32 3733386 0% /tmp /dev/da0s1f 9499186 2032248 6707004 23% /usr /dev/da1s1a 68825650 10008236 53311362 16% /opt 115572 : mount /dev/da0s1a on / (ufs, local) devfs on /dev (devfs, local) procfs on /proc (procfs, local) /dev/da0s1d on /var (ufs, local, soft-updates) /dev/da0s1e on /tmp (ufs, local, soft-updates) /dev/da0s1f on /usr (ufs, local, soft-updates) /dev/da1s1a on /opt (ufs, local, soft-updates) pid906@rx300:/snap on /snap (nfs) 115569 : snapshot list Filesystem User User% Snap Snap% Snapshot / 181MB 36.5% 3MB 0.7% daily.0 / 181MB 36.5% 3MB 0.7% weekly.0 /var 2377MB 15.0% 40MB 0.3% daily.0 /var 2377MB 15.0% 9MB 0.1% hourly.0 /var 2377MB 15.0% 10MB 0.1% hourly.1 /var 2377MB 15.0% 10MB 0.1% hourly.2 /var 2377MB 15.0% 10MB 0.1% hourly.3 /var 2377MB 15.0% 10MB 0.1% hourly.4 /var 2377MB 15.0% 10MB 0.1% hourly.5 /var 2377MB 15.0% 10MB 0.1% hourly.6 /var 2377MB 15.0% 10MB 0.1% hourly.7 /var 2377MB 15.0% 44MB 0.3% weekly.0 /usr 1984MB 21.4% 61MB 0.7% daily.1 /usr 1984MB 21.4% 67MB 0.7% daily.2 /usr 1984MB 21.4% 11MB 0.1% weekly.0 /opt 9773MB 14.5% 361MB 0.5% daily.0 /opt 9773MB 14.5% 757MB 1.1% daily.2 /opt 9773MB 14.5% 142MB 0.2% hourly.0 /opt 9773MB 14.5% 156MB 0.2% hourly.1 /opt 9773MB 14.5% 171MB 0.3% hourly.2 /opt 9773MB 14.5% 187MB 0.3% hourly.3 /opt 9773MB 14.5% 204MB 0.3% hourly.4 /opt 9773MB 14.5% 218MB 0.3% hourly.5 /opt 9773MB 14.5% 231MB 0.3% hourly.6 /opt 9773MB 14.5% 717MB 1.1% weekly.0 Fix: Don't know! How-To-Repeat: Keep periodically snapshots. Perhaps the creation of files on the snapshotted filesystem is a prerequisite to trigger the crash (compare PR kern/68757).
Just another crash occurred, during manipulation of an snapshotted file system. Taken of newly snapshots are disables since yesterday. Output of info file: Dump header from device /dev/da0s1b Architecture: i386 Architecture Version: 2 Dump Length: 1073242112B (1023 MB) Blocksize: 512 Dumptime: Tue Aug 21 10:42:04 2007 Hostname: rx300.hrz.uni-bielefeld.de Magic: FreeBSD Kernel Dump Version String: FreeBSD 6.1-RELEASE-p12 #0: Tue Jan 16 16:58:07 CET 2007 root@rx300.hrz.uni-bielefeld.de:/opt/tmp/obj/usr/src/sys/PMX Panic String: snapblkfree: inconsistent block type Dump Parity: 1638409495 Bounds: 15 Dump Status: good Anyone interested in this problem? Best regards Lars -- Dr. Lars Köller, Raum: V0-318, Tel: +49 521/106-4964, FAX: -154964 http://www.uni-bielefeld.de/hrz/raum/plan.htm
Hi, why is kern/68757 still critical (submitted 2004), and our case only serious? I know release mgt of 7.0 costs a lot of time, but snapshot functionality hasn't reached production quality yet. I'm really amezed about that cause snapshots are introduced in 5.0. Perhaps it's a good idea to fix this problem before 7.0 release is out? Many thanks and regards Lars
Responsible Changed From-To: freebsd-bugs->freebsd-s Over to maintainer(s).
Responsible Changed From-To: freebsd-s->freebsd-fs fix typo
State Changed From-To: open->feedback Can you still reproduce this PR with a more recent 6.X or 7.X release? Please note that the RELENG_6 went EoL a few weeks ago.
State Changed From-To: feedback->closed Feedback timeout.