Nick Lunt wrote:
Here's my plan of action. 1. install gfs-utils (don't install clustering or "cluster storage"). 2. present the LUNs to each server. 3. do lvm stuff on each server. 3. mkfs.gfs2 -j5 -p lock_dlm -t what:goeshere /dev/vg02/lvolx . 4. mount from each server and put in fstab on each server. My problem is at number 3. If I want to use lock_dlm I need to specify a table <clustername:filesystemID>. However I do not want to setup a cluster I just want GFS.
You don't have a choice. Clustering isn't optional. GFS will not work without clustering because DLM depends on it to establish quorum and suchlike.
I can use lock_nolock which does not require a table but will that render the filesystem useless with >1 server having access to it ?
That will corrupt the FS because locking won't work.
Unfortunately I've found no documentation on a GFS only setup, I would be very grateful if somebody could help me out.
The reason there's no documentation is because it's not possible. What's the problem with having cluster running, though? It only takes a few lines of XML in cluster.conf, and this is reasonably well documented. You don't need any failover services, just the node entries.
You could use OCFS2 instead, which doesn't require RHCS, but you will still need to set up a roughly equivalent configuration to get it's own locking working, so it still wouldn't save you any effort.
Gordan -- Linux-cluster mailing list Linux-cluster@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/linux-cluster
