Heartbit + Pacemaker + drbd error
1.
============
2.
Last updated: Tue Oct 12 22:00:02 2010
3.
Stack: Heartbeat
4.
Current DC: ha2 (8de4b285-24ef-4d59-9417-e7a137584e3a) - partition with quorum
5.
Version: 1.0.8-042548a451fce8400660f6031f4da6f0223dd5dd
6.
2 Nodes configured, 1 expected votes
7.
2 Resources configured.
8.
============
9.
10.
Online: [ ha1 ha2 ]
11.
12.
Resource Group: WebServices
13.
fs_webfs (ocf:
14.
Master/Slave Set: msdrbdwebfs
15.
drbd_webfs:1 (ocf::linbit:drbd): Slave ha2 (unmanaged) FAILED
16.
Masters: [ ha1 ]
17.
18.
Failed actions:
19.
drbdwebfs:1start_0 (node=ha2, call=4, rc=5, status=complete): not installed
20.
drbdwebfs:1stop_0 (node=ha2, call=7, rc=5, status=complete): not installed
2 Replies
You can create some really sophisticated setups to failover, and sometimes it works. But a lot of the times the failover process for me got stuck because of open files or some other reason and getting it unstuck was a huge pain. Heartbeat v2 supposedly added some great features, but in practice (at least for me) it added a lot of needless complexity, and the documentation is horribly lacking.
Best of luck getting it working. Perhaps I just didn't have something setup correctly, or maybe there have been improvements since two years ago when I was using it, but that was just my experience FWIW
Check /proc/drbd on both machines. Also dump your pacemaker configuration (crm configure)
If you're playing with dual primaries you should really look into a filesystem with distributed locking mechanisms, like GFS+ or OCFS2. I prefer the latter one, for no apparent reason!