On Tue, Dec 29, 2009 at 11:23 PM, Gordan Bobic <gordan@xxxxxxxxxx> wrote:
Well, the good thing about being in a RH class is that you can do all kinds of sick, twisted evil things just to see what happens. I've also made the mistake of doing things like not changing the locking_type in lvm.conf to 3 and forgetting to start clvmd. Any of those can lead to strange and exciting times with GFS.
We're fine with GFS where we are. I've done some benchmarking on GFS2 and it's performance didn't come anywhere near what we could do with GFS.
Ian Hayes wrote:I'm assuming you mean that you were mkfs-ing one and then trying to mount the other. I'm vehemently against putting everything on lvm just for the sake of it, but I've never had a problem with mkfs-ing or mount-ing either, as long as it's consistent. I tend not to partition iSCSI and DRBD volumes, so I know that working direct with the whole block device works just fine.
I had a similar problem in my Redhat Clustering and Storage Management class the other week. I believe the problem was with a couple of mistakes I made while playing around in one of the labs. I know once it was because I was trying to mount the block device instead of the logical volume.
Well, the good thing about being in a RH class is that you can do all kinds of sick, twisted evil things just to see what happens. I've also made the mistake of doing things like not changing the locking_type in lvm.conf to 3 and forgetting to start clvmd. Any of those can lead to strange and exciting times with GFS.
Don't quite me on this, but I'm pretty sure GFS2 is deemed stable as of RHEL 5.4 (or was it 5.3?). Having said that, I haven't yet deployed any GFS2 volumes in production, and don't plan on doing so imminently, so draw whatever conclusions you see fit from that. ;)IIRC, gfs2 is still under development and considered experimental. There's tons of documentation for production-quality GFS and I imagine once gfs2 gets more mainlined, this will be the case also.
We're fine with GFS where we are. I've done some benchmarking on GFS2 and it's performance didn't come anywhere near what we could do with GFS.
-- Linux-cluster mailing list Linux-cluster@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/linux-cluster
