Hi. After learning about GFS and its locking, I'm having trouble answering a question: What would be the problem with this scenario? node 1: create a RAID5 on three shared disks, /dev/md0 create a GFS filesystem on /dev/md0 node 2: using the same config file, start /dev/md0 with the same shared disks node 3: using the same config file, start /dev/md0 with the same shared disks Now everybody joins the cluster and mounts the filesystem. It seems like the GFS locking would take care of most of the problems I can think of, but I suspect there may be something tricky. Should this work? -- Ed L Cashin <ecashin@xxxxxxxxxx>