Messages in this thread | | | Date | Wed, 10 Jun 2009 18:42:16 +0200 | From | Harald Dunkel <> | Subject | Re: how to stop kernel nfsd? |
| |
Christian Kujau wrote: > On Mon, 8 Jun 2009, harald.dunkel@t-online.de wrote: >> Which signal would be correct to gracefully kill the kernel nfs daemons? > > What are you trying to achieve? "killing", "terminating" or "graceful > reloading", as in "re-exporting the shares", which can be done with > exportfs(8). >
Its a HA setup. Heartbeat is supposed to terminate the nfs service, umount the local filesystem, set the drbd resources to "secondary", and release the shared IP address (in this sequence).
To terminate the nfs service Lenny's /etc/init.d/nfs-kernel-server tries
kill -INT <pid>
for all nfs daemons. Even though nfs is not active the nfs daemons don't go away and keep the local mount point busy. If I patch the nfs runlevel script to use SIGQUIT or SIGHUP instead, then the nfs daemons terminate.
Lenny's kernel is 2.6.26 plus patches.
Surely I am not asking for support for Lenny's kernel. But it would be interesting to know what the kernel nfs daemons are supposed to do when they receive a signal. AFAICS this is not documented.
Regards
Harri
[unhandled content-type:application/pgp-signature] | |