[Linux-cluster] Duplicate Volumes, GFS fail over

[Greg Swift]

isplist at logicore.net wrote:
> One of the reasons I went with fibre channel was because of the redundancy it 
> can give me as well as the centralized storage.
> There is one aspect of this I've not found any solutions to yet however in 
> dealing with GFS, redundant paths.
>
> As you can see, I have 4 storage devices, three of which have dual paths in 
> case of failure. The problem is that the system thinks this is duplicate and 
> possibly an error as it sees it, I'm not sure. 
>
> I thought I would ask before trying to put the dual paths to use. Here is an 
> output;
>
>   Found duplicate PV TB3VUn1m3CBOj8dRnRjAbRuD3ZIyBp7t: using /dev/sde1 not 
> /dev/sda1
>   Found duplicate PV uXjYfhj4NlvQphf1DIz28psAAkFmJRaM: using /dev/sdf1 not 
> /dev/sdb1
>   Found duplicate PV 3mlyNROBtWp4a3LXQEKBiCwk3pJear7u: using /dev/sdg1 not 
> /dev/sdc1
>   ACTIVE            '/dev/VolGroup01/rimfire' [572.72 GB] inherit
>   ACTIVE            '/dev/VolGroup04/web' [318.85 GB] inherit
>   ACTIVE            '/dev/VolGroup03/qm' [745.76 GB] inherit
>   ACTIVE            '/dev/VolGroup02/sql' [745.76 GB] inherit
>
> I don't have a dual path on the VolGroup01 device because it's just manual 
> storage I use here and then.
>
> The question is; How should I deal with the dual paths and how do I set up 
> volumes/GFS to be fault tolerant in this respect. If one path fails, I'll want 
> GFS to fail over to the second path to the same storage.
>
> Mike
>   
use the device-mapper-multipath package You would do it beore you create 
your lvm volumes. Depending on the equipment it can be as easy as just 
removing the blacklist in the /etc/multipath.conf, but sometimes it 
might take contacting your vendor (like i had to w/ emc) to find the 
best setup to run the multipath cleanly.

-greg

-- 
http://www.gvtc.com
--
“While it is possible to change without improving, it is impossible to improve without changing.” -anonymous

“only he who attempts the absurd can achieve the impossible.” -anonymous