make_recovery delay

[Ryan Ma]

When NFS is not responding, make_recovery do not start unit NFS is ready.

Is there any way I can ignore NFS not responding?

[Bill Hassell]

This is a 'feature' of NFS. A mounted filesystem from an NFS location is treated the same as a local disk, that is, loss of the disk breaks just about everything (bdf, login, make_recovery, etc).

The only to avoid this is unmount all NFS mountpoints before doing important operations like make_recovery. If NFS servers are not stable, I would not recommend using them at all, especially in a production environment.

[Ryan Ma]

Sometimes NFS not responding is due to NFS server reboot, etc. When NFS is not responding, you can't even umount it, right?

So it is a little trouble.

[Hooi Siew Hoong_1]

Hi,

Wait for the NFS server to boot up and then unmount your NFS filesystem. But sometimes, even after the NFS server is up, the mount is still stale, so I guess the best way now is to comment all the nfs entries in /etc/fstab and reboot.

In my enviroment, we always umount all exported FS when we reboot any NFS server, just as a pre-caution.

cheers.

[Michael Tully]

Hi,

As pointed out the best bet is to avoid using
any type of NFS whilst cutting ignite tapes.

I avoid NFS on production servers like the
plague. NFS has never been completely stable
due to it always relying upon the other side
to be available.

-Michael