[Linux-cluster] NFS making umount fail.

[Matthew Geier]

  My cluster won't shutdown properly and fails the service unless I 
manually kill of nfsd first

Jan 23 17:30:39 Aristotle0 clurgmgrd: [8508]: <info> Stopping Samba 
instance "SMB-Export"
Jan 23 17:30:39 Aristotle0 clurgmgrd: [8508]: <info> unmounting /home
Jan 23 17:30:39 Aristotle0 clurgmgrd: [8508]: <notice> Forcefully 
unmounting /home
Jan 23 17:30:40 Aristotle0 clurgmgrd: [8508]: <warning> Dropping 
node-wide NFS locks
Jan 23 17:30:50 Aristotle0 clurgmgrd: [8508]: <info> unmounting /home
Jan 23 17:30:50 Aristotle0 clurgmgrd: [8508]: <notice> Forcefully 
unmounting /home
Jan 23 17:30:51 Aristotle0 clurgmgrd: [8508]: <info> Sending reclaim 
notifications via Aristotle0.arts.usyd.edu.au
Jan 23 17:30:51 Aristotle0 rpc.statd[5097]: Version 1.0.6 Starting
Jan 23 17:30:51 Aristotle0 rpc.statd[5097]: Flags: No-Daemon Notify-Only
Jan 23 17:30:51 Aristotle0 rpc.statd[5097]: unlink 
(/tmp/statd-Aristotle0.arts.usyd.edu.au.4919/sm.bak/129.78.16.20): 
Permission denied
Jan 23 17:30:54 Aristotle0 rpc.statd[5097]: Caught signal 15, 
un-registering and exiting.
Jan 23 17:30:54 Aristotle0 clurgmgrd: [8508]: <err> 'umount /home' 
failed, error=0
Jan 23 17:30:54 Aristotle0 clurgmgrd[8508]: <notice> stop on fs "Files - 
home" returned 2 (invalid argument(s))

  At which point the service is failed. It will not shutdown until I 
/etc/init.d/nfs stop first.


  I also use quota's, and unfortunately, 'quotaon -a' is run in the 
system init scripts BEFORE the filesystems are mounted as they are 
mounted by the cluster service.

  As a result I lose enforement of quotas all the time....