Saturday morning live: SLES9+DRBD upgrade
I have not worked on saturday mornings (at least officially) since 2001, but today I planned an upgrade of a SLES 9 cluster to the latest patchlevel. The cluster uses drbd to replicate two disks, used mainly for samba sharing.
My plan was to try a rolling upgrade, even tough nobody is at work today, so that I could get some experience on this and avoid other saturday mornings live ;-) . From what I understand drbd can be upgraded on one node at the time as the protocol is compatible between minor versions and since the upgrade was from 0.7.pre to 0.7.5 I thought: ok, I can do it.
Believe it or not it is not true: the pre thing in the original SLES9 is not compatibile with the one that comes from online update. So I had to upgrade both and then reboot. The whole process took about one hour (add some time for safety checks on partitions, etc) and I was really disappointed that even in the suse support portal this was not mentioned at all!
Except for that drbd did not force a full resync (as it would have done in the 0.6.x versions) and the cluster came up very quickly.
Have a nice weekend!
My plan was to try a rolling upgrade, even tough nobody is at work today, so that I could get some experience on this and avoid other saturday mornings live ;-) . From what I understand drbd can be upgraded on one node at the time as the protocol is compatible between minor versions and since the upgrade was from 0.7.pre to 0.7.5 I thought: ok, I can do it.
Believe it or not it is not true: the pre thing in the original SLES9 is not compatibile with the one that comes from online update. So I had to upgrade both and then reboot. The whole process took about one hour (add some time for safety checks on partitions, etc) and I was really disappointed that even in the suse support portal this was not mentioned at all!
Except for that drbd did not force a full resync (as it would have done in the 0.6.x versions) and the cluster came up very quickly.
Have a nice weekend!