Hello,
I'm currently fighting a very obscur problem: when the kernel nfs-server is
started from heartheat, it can't be stopped the usual way from the
debian-nfs-kernel-server script.
As long as I manually run '/etc/init.d/nfs-kernel-server start', stopping
works by running '/etc/init.d/nfs-kernel-server stop'. But when the same
commands are called from heartbeat, the nfsd-processes simply won't be
killed. Running '/etc/ini.d/nfs-kernel-server stop' manually also doesn't
work then.
Well, from the debian script the nfsd is killed with signal 2, killing it
manually with signal 1 or signal 9 works, though sometimes the kernel will
log "rpciod: active tasks at shutdown?!".
Any ideas what might be the differences between starting nfs from heartbeat
and starting it as root? I already tried to trace this down and first I
thought heartbeat might start the services in the wrong order, but finally I
created an absolut identical environment so that heartbeat only starts and
tries to stop nfs.
Is signal 2 anyway sufficient to kill the nfsd's? From the Suse-7.3 nfsserver
script I see that those daemons are killed there by default with
signal KILL (9).
System configuration:
- linux-2.4.26 on dual opteron
- debian woody + several packages from sarge
- nfs-utils-1.0.6-3
Thanks for any help,
Bernd