On 08/31/2015 02:41 AM, Ulrich Windl wrote: > Actually when using ext3 I wouldn't use cLVM, so this variant > wouldn't exist then.
Hi Ulrich, I tried it first without cLVM (so I could get rid of DLM as well) but then you have to modify /etc/lvm/lvm.conf accordingly. You have to activate the "volume_list" option and exclude from it the VGs that are going to be used on the cluster. If you don't activate the option (which is commented by default) the LVM resource agent is going to complain (it's just a precaution). As a consequence of enabling that option you're not allowed to manually perform operations on the VG (lvextend etc) while the cluster is running. In the end it turns out using cLVM was easier.... > I see the danger that someone might mount the ext3-LV in a second > node by mistake. Sometimes this will corrupt the filesystem in a way > that fsck cannot repair, and maybe you'll discover that days after it > happened... Now with cLVM....when using the LVM resource agent (which activates the VG) there's an option to activate it *exclusively*; therefore the VG is only active on the node where the LVM resource is running. On the other node the VG isn't activated and therefore you won't even have the logical-volume device-files created (to even try to mount it). You may manually perform "vgchange -a y myVG" (in order to try to activate it and create the device-files) to active it but it fails to do so (wisely). Regards, Jorge _______________________________________________ Users mailing list: Users@clusterlabs.org http://clusterlabs.org/mailman/listinfo/users Project Home: http://www.clusterlabs.org Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf Bugs: http://bugs.clusterlabs.org