Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
I've just set up DRBD 8 (8.3) on a KVM host with the secondary (for the moment) being an old server, just to make sure my KVM images are secured against primary failure. At the moment I'm not interested in clustering. What I've done is set up a nested LVM, so that each KVM guest has its own logical volume, and that way I can eventually load balance (when I reconfigure my other KVM server to use drbd). As an added wrinkle, because these guests are being used to work on some confidential data, we are contractually required to have everything encrypted, so the main LVM is encrypted with LUKS. I actually have the set up working and one KVM guest is up and running and working great. My questions surround automating the restart process. Obviously, on reboot (or if the encrypted LVM is unmounted), someone will have to enter a password, but I can script for that pretty easily. What I'm more concerned about is bringing up the DRBD array properly, and failing gracefully. This is the barebones scripts I have so far On the primary: #!/bin/sh if ! cryptsetup luksOpen /dev/mapper/kvmserver02-guestscontainer vg-guests; then echo "Invalid password"; exit 1; fi vgchange -a y vg-guests modprobe drbd service drbd start drbdadm primary drbd-guest1 drbdadm primary drbd-guest2 mount /dev/drbd/by-res/drbd-guest1 /var/lib/libvirt/images/guest1 mount /dev/drbd/by-res/drbd-guest2 /var/lib/libvirt/images/guest2 On the secondary: #!/bin/sh if ! cryptsetup luksOpen /dev/mapper/backupserver-guestscontainer vg-guests; then echo "Invalid password"; exit 1; fi vgchange -a y vg-guests modprobe drbd service drbd start drbdadm secondary drbd-guest1 drbdadm secondary drbd-guest2 This seems lacking to me. What if the backup server doesn't come on line? How do I force the primary online if that happens? -- Aaron Clausen mightymartianca at gmail.com