Hi,
I got two responses on this. Thanks to
Robert Mulley <genius_at_one.net.au>
Sakellaris Alexander <asakelaris_at_cosmote.gr>
They recommend to use the stop script to check for users/open files in
that mount point and then kill them.
I'm not sure what actually should be checked as login is disabled and
no process will use that service on A. A just functions as the nfs server
for this service.
Any other ideas are welcome. Thanks anyway.
Regards,
Raymond
On Fri, Jan 29, 1999 at 05:15:43PM +0800, Pang Wai Man Raymond wrote:
> Hi Manager,
>
> We encounter inconsistant asedb after one of the member rebooted. Asemgr
> shows one of the nfs service is on its flavored member B while it is
> actually on A. This is the second time it happened. Our experience tells
> us that all members will crash if this service is being relocated.
>
> Configuration:
> DU4.0B + TCR1.4A
> 3 x 4100 A: Flavored member of some nfs services
> B: Flavored member of some nfs services +
> the problematic services
> C: Not a flavored member of any nfs services
>
> Sequence of the problem occurs:
> 1. B,C shutdown
> 2. all services failed over to A
> 3. B boot up.
> 4. daemon.log on A shows the service is busy and cannot be relocated.
> 5. asemgr shows that it is relocated to B (the flavor member).
>
> daemon.log
> ==========
> Jan 23 13:01:26 sysA ASE: mcsysA Agent Error:
> /var/ase/sbin/ase_mount_action: /usr/var/ase/mnt/nfs_he_info_ipoc/info_filter_mnt: Device busy
> Jan 23 13:01:26 sysA last message repeated 9 times
> Jan 23 13:01:26 sysA ASE: mcsysA Agent Error: /var/ase/sbin/ase_mount_action: Unable to umount /var/ase/mnt/nfs_he_info_ipoc/info_filter_mnt
>
> Although I know that this service is always being heavily loaded by
> clients, it shouldn't be so bad for cluster.
>
> Is there anything we could do to prevent this from happening again.
>
> TIA.
>
> Regards,
> Raymond
Received on Tue Feb 09 1999 - 06:12:28 NZDT