Not the case here but there are some hosts in rmttab that have been 'inop' for months as we consolidated their function and turned the box off.
-----Original Message----- From: tmac [mailto:tmacmd@gmail.com] Sent: Friday, April 20, 2007 3:29 PM To: Brian Dunbar Cc: toasters@mathworks.com Subject: Re: ndmpcopy and the newb Part II
If that machine disconnected un-cleanly from the NetApp... i.e. It did not tell the NetApp that it was unmounting it, Then the filer will think (through the rmtab and other files) that some host or hosts still have things mounted, when in fact, they do not.
--tmac
On 4/20/07, Brian Dunbar Brian.Dunbar@plexus.com wrote:
Me again.
When last we met I had a question about ndmpcopy - the copy went over well - files moved, the application that needs those files had a brief outage and all is well.
I'm cleaning up the source volume - 'best practice' here is to remove files, remove snapshots and shrink the volume to 10g should we need a place to stash files.
Removed the files, done. Removed the snapshots, done.
Except for one guy that is being stubborn. The fellow with just a bit more experience with these than I do claims it must be because a
server
is still has that volume mounted and if I showmount that appears to be the case .. except the remote host in question no longer has that mounted - going by 'df -k' and /etc/vfstab.
Now - a very long time ago - back in November of 06 - this volume was cloned and mounted so that an application under development could see the contents of the file store there.
Is there something special - clearly there might be - about clones and mirrors that require an extra effort to delete a snapshot? Or am I doing something fundamentally wrong?
I would not ordinarily go the newb well twice in a fortnight but I've been looking at this since early this morning and I'm no closer to figuring out WHY than when I started.
Thanks.
HOST:/:# rsh co-dsk-007 snap list aprod4 Volume aprod4 working...
%/used %/total date name
28% (28%) 0% ( 0%) Nov 16 15:30 clone_aprod4_cln_aglif.1 (busy,vclone)
"Busy? Surely the OS isn't going to let me blow away something that
is
marked as busy .."
HOST:/:# rsh co-dsk-007 snap delete aprod4 clone_aprod4_cln_aglif.1 Snapshot clone_aprod4_cln_aglif.1 is busy because of
"How bout that. What about the nfs mounts?"
LOCALHOST:/sbin:showmount -a co-dsk-007 | grep aprod4 206.209.246.192:/vol/aprod4 co-ap-103v.plexus.com:/vol/aprod4 co-ap-930v.plexus.com:/vol/aprod4_cln_aglifs
"Both hosts have been vi /etc/vfstab 'd and umounted and even rebooted to death at this point .."
Brian Dunbar Systems Administrator II Plexus
brian.dunbar@plexus.com Desk: (920) 751-3364 Cell: (920) 716-2027