[DRBD-user] Warning: DRBD/heartbeat killed by korrupt fuser (21.8) on Debian Sarge

Dr. Volker Jaenisch volker.jaenisch at inqbus.de
Fri Dec 9 01:08:19 CET 2005

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


Hello DRBD-Users

Setup
Stock Debian Sarge (stable): Kernel 2.6.12-1-686-smp (testing, the only 
testing thing on the machine). DRBD 0.7.11-1 (Debian way compiled 
module). fuser 21.8. P4 HT Machine.

We encounter problems with heartbeat steering drbd.
Problem:
drbd/heartbeat reboots the released node since it could not unmount the 
drbd-device.
Logfile:
Dec  4 11:54:09 mail2 heartbeat: ERROR: Return code 1 from 
/etc/ha.d/resource.d/Filesystem
Dec  4 11:54:10 mail2 heartbeat: info: Retrying failed stop operation 
[Filesystem::/dev/drbd0::/shared::ext3]
Dec  4 11:54:10 mail2 heartbeat: info: Running 
/etc/ha.d/resource.d/Filesystem /dev/drbd0 /shared ext3 stop
Dec  4 11:54:10 mail2 heartbeat: debug: Starting 
/etc/ha.d/resource.d/Filesystem /dev/drbd0 /shared ext3 stop
Dec  4 11:54:10 mail2 heartbeat: ERROR: Couldn't unmount /shared
Dec  4 11:54:10 mail2 heartbeat: debug: /etc/ha.d/resource.d/Filesystem 
/dev/drbd0 /shared ext3 stop done. RC=
1
Dec  4 11:54:10 mail2 heartbeat: ERROR: Return code 1 from 
/etc/ha.d/resource.d/Filesystem
Dec  4 11:54:10 mail2 heartbeat: CRIT: Resource STOP failure. Reboot 
required!
Dec  4 11:54:10 mail2 heartbeat: CRIT: Killing heartbeat ungracefully!
Dec  4 11:55:16 mail2 syslogd 1.4.1#17: restart.

The problem was easily tracked down to the call to "fuser" in the widely 
used file
/etc/ha.d/resource.d/Filesystem
...
    # Kill all processes open on filesystem
    $FUSER -km $MOUNTPOINT
...

which does simply not what it should. The processes sitting on our 
DRBD-Device mounted to /shared were _NOT_ killed.

The weird phaenomenon is

fuser -mk        -> does not work
fuser -m -k      -> works
fuser -km         -> works

This outcome eliminates any machine/setup dependency and targets the 
error explicitly to the input-parameter evaluation of the fuser program.

I will write a bugreport to the fuser-people, too.
I posted this here since some of the DRBD threads (e.g. "NFS did not 
release share") may be related to that issue.

Best regards,

Volker Jaenisch

-- 
====================================================
   inqbus it-consulting      +49 ( 341 )  5643800
   Dr.  Volker Jaenisch      http://www.inqbus.de
   Herloßsohnstr.    12      0 4 1 5 5    Leipzig
   N  O  T -  F Ä L L E      +49 ( 170 )  3113748
====================================================




More information about the drbd-user mailing list