It's me again with a different problem.
I'm running 188.8.131.52 and keepalived 1.1.20, with two alternative
interfaces configured for ipvs sync operation, eth0.1 and eth1.3
While hunting for does-not-sync problems (other mail of yesterday +
today) I probably got confused by the master/backup partner deciding to
use a different interface on each side.
Normally the both use the same one (as determined by kernel logging
output when the sync threads start). But every now and then each one
tells me it's using a different interface than the other.
This morning I reached that state when I stopped the keepalived on the
master. The backup became master. Then I restarted the keepalived on the
former master, it became master again with the sync thread on eth1.3,
but then the backup server switched to backup and had the sync thread
running on eth0.1.
When in that situation, no syncing takes place at all.
I'm a bit unsure who is responsible for that, the ipvs kernel code, or
the keepalived. Does anybody know?
Please read the documentation before posting - it's available at:
LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
or go to http://lists.graemef.net/mailman/listinfo/lvs-users