Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
> So what is the best way to partition tihs beast? And why? > Well, that depends on what you want to do. First off, I think booting off a drbd isn't such a hot idea. Two clustered systems are never really identical. At the very least, hostnames and IPs and sshd keys are different. If you do boot off of a drbd, you'll need the drbd module in your ramdisk. I bet most modern Debian/RedHat-ish systems would figure this out for you at mkinitrd time. Now, whether it's better to put LVM on top of drbd or the other way around? That really depends on your need. As an example, I'm testing this low-budget setup right now: Four disks LVM'd together using LVM striping with 22 equal sized LVs. Each LV has a drbd resource on top of it in primary/secondary. I then format and mount the dbrds. Why did I do that? I have 22 logically separate things I want to manage. I want to be able to switch primary per 'thing', rather than the entire set at once. Because the 22 drbds are all really backed by a single raid0 (LVM provided, not md). If one box looses a disk all 22 drbds on that box detach from local disk and read from the network, but I didn't do 22 drbds to get individual reliability. I did them to get the granularity I wanted to administratively migrate each of my 22 things one by one if I wanted.