Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Hello, On Tue, 11 Jan 2005 14:40:38 +0100, Laurent Simonneau <liste at alixen.fr> wrote: > After a successfull drbd synchro, I try to run > mke2fs -j /dev/drbd0 ... and drbd crash ! The connexion > with secondary node is broken. This happened to me with drbd 0.7.7 and kernel 2.4.2[678]. > It seems to work with protocol A, but I don't really understand why ... Did not try that - but I dont want to run protocol A anyhow :-) > But is it really interesting to use DRBD over LVM ? I use LVM for > snapshots to have more consistant backup. I saw on mailing list, > FAQ, ... a lot of contradictory informations about DRBD and LVM > snapshot. > Does it works or not ? I have working drbd-over-lvm2 setups on a production servers, working perfectly with protocol C. I'm running linux 2.6.9, drbd 0.7.7 and also experimental vserver patches. The question of using LVM underneath is a bit philosophical - I like the flexilibity it gives me, especially since the standby machine is linked with four other machines. To add another one machine, I'd just lvcreate a new logical volume, and configure drbd on top of this. Hopefully not all primarys will fail at the same time, or the standby will get a lot to do =). One server is scrapped? No problem, remove drbd and lvremove it - the space can then efficiently be used for another server. Just the other day, I needed more space for one vserver. Stop the drbd, lvresize the logical volume on both nodes, drbdadm resize, start drbd, resize2fs /dev/drbd0, mount and start the vserver. It would be _a lot_ more work if regular partitions had been used. But I guess it all depends on your needs and the setup. Regards