[Linux-cluster] gfs for exsisting disks

[Digimer]

On 11/14/2012 09:54 AM, getridofthespam wrote:
> Hi all,
> 
> I have a Centos 6.3 with a SAN storage attached. mount extract:
> 
> /dev/mapper/mpathcp1 on /3parslice1 type ext4 (rw)
> /dev/mapper/mpathbp1 on /3parslice2 type ext4 (rw)
> 
> The slices are ext4 formatted.
> 
> Now I want to add a second server that needs to access the
> same disk slices.
> 
> Is gfs the solution? Can I keep the data on the disks?
> Any procedure to follow available?
> 
> Tnx for all answers.

Unless there is some voodoo I don't know about, no, you will need to
backup, reformat gfs2 and restore the files.

Yes, gfs2 will allow 2+ nodes to access the same data on the SAN, but
there are considerations to be aware of. First is that the distributed
locking (dlm) comes at an overhead cost. before each write can happen, a
lock must be requested from the cluster. If you have disk intensive
apps, this might cause unacceptable delays.

Also, you *must must must* have testing, working fencing for gfs2 to be
safe.

So it might be worth putting together a test case before you commit to
converting production boxes, if you have the resources.

-- 
Digimer
Papers and Projects: https://alteeve.ca/w/
What if the cure for cancer is trapped in the mind of a person without
access to education?