[DRBD-user] dopd failover

Rois Cannon rois at cobiz.com
Fri Dec 14 00:44:00 CET 2007

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


I changed drbdmeta and drbdsetup to the recommended permissions but
changed the group to hacluster (instead of haclient)  Now looks like
this:
-rwsr-xr-- 1 root hacluster 47840 2007-09-06 04:05 /usr/sbin/drbdmeta*
-rwsr-xr-- 1 root hacluster 33804 2007-09-06 04:05 /usr/sbin/drbdsetup*

I then tested dopd by disconnecting the drbd data path (eth1).  dopd did
exactly what it was suppose to.  node2 had it's resource magically
outdated.  Yeah!!!!!  

btw . . . I turned off the ping on eth1 so heartbeat wouldn't be doing
anything while I was testing this.  Looking at the big picture I'm
guessing I should should not have ping running on this nic and just do a
crossover cable for the data path???  I was kind of hoping to use this
path as a private gigabit lan for some other data transfers as well for
which I would need a VIP.  Better to add a 3rd nic????? 

Now that bad news . . .

if node1 (as primary) is powered off I'm still getting the following
error:
Dec 13 15:21:24 svr92 ResourceManager[22764]:
debug: /etc/ha.d/resource.d/drbddisk home start done. RC=20
Dec 13 15:21:24 svr92 ResourceManager[22764]: ERROR: Return code 20
from /etc/ha.d/resource.d/drbddisk
Dec 13 15:21:24 svr92 ResourceManager[22764]: CRIT: Giving up resources
due to failure of drbddisk::home

and node2 (svr92) never becomes primary.

I'm guessing this isn't a dopd problem but a fencing problem.
If I leave fencing and dopd disabled it rolls over just fine but
of coarse when node1 comes back on-line it is NOT outdated.

Thoughts???




More information about the drbd-user mailing list