[Linux-cluster] Clvm over gnbd + rgmanager

Ion Alberdi ialberdi at histor.fr
Mon Mar 7 16:39:25 UTC 2005


Derek Anderson wrote:

>
> This is indeed a problem with fs.sh.  It attempts to umount with the 
> device name rather than the mount point, which is not guaranteed to 
> work in all situations.  From mount(8):
>
> "A file system is specified by giving the  directory  where
>       it  has been mounted. Giving the special device on which the 
> file sys-
>       tem lives may also work, but is obsolete, mainly because it 
> will  fail
>       in case this device was mounted on more than one directory."
>
> So that's why it is failing.  The solution is probably to umount with 
> the mount point.  I'll open a bug and let Lon decide.
>
I modified fs.sh,
umount $dev -> umount $mp line 821,
and I had that on syslog
Mar  7 17:08:04 gump kernel: kjournald starting.  Commit interval 5 seconds
Mar  7 17:08:04 gump kernel: EXT3 FS on dm-0, internal journal
Mar  7 17:08:04 gump kernel: EXT3-fs: mounted filesystem with ordered 
data mode.
Mar  7 17:08:05 gump clurgmgrd[7356]: <notice> Resource group hello started
Mar  7 17:09:02 gump clurgmgrd[7356]: <info> Magma Event: Membership Change
Mar  7 17:09:02 gump clurgmgrd[7356]: <info> State change: buba UP
Mar  7 17:09:05 gump clurgmgrd[7356]: <notice> status on fs "my fs" 
returned 1 (generic error)
Mar  7 17:09:05 gump clurgmgrd[7356]: <notice> Stopping resource group hello
Mar  7 17:09:05 gump clurgmgrd[7356]: <notice> stop on script "Hello 
Script" returned 1 (generic error)
Mar  7 17:09:05 gump clurgmgrd[7356]: <crit> #12: RG hello failed to 
stop; intervention required
Mar  7 17:09:05 gump clurgmgrd[7356]: <notice> Resource group hello is 
failed

The file system remains mounted after all this logs...

I think the strangest thing doesn't come from stop, but from status:
Mar  7 17:09:05 gump clurgmgrd[7356]: <notice> status on fs "my fs" 
returned 1 (generic error)
I don't understand why this message appears some minutes after launching 
the service and using a logical volume ,whereas the status doesnt change 
with  other devices (IDE, gnbd, file devices (I used mount -o loop)).

> Thank you for pointing this out.
>
You're welcome, I'd be glad if I could help.




More information about the Linux-cluster mailing list