Hello *,
I may be missing something obvious, but we had an incident this morning which
seemed extremely strange.
Basically, we have a NAT lvs cluster with 2 different clusters on it, both web,
both of them having 3 real servers that traffic is distributed to. Things have
been running great.
This morning, I rebooted one of the real servers that belong to the first
cluster.
Suddenly, all of lvs died.
The logs tell me that "nanny process on RealServer3/Cluster1 died, shutting down
lvs".
Does this make any sense to anyone? Why would cleanly rebooting one of the
realservers cause all of lvs to shut down.
Is this expected behavior? Is my config possibly fubared? Is this a bug?
(I'm using ipvs for kernel 2.2.17 and Red Hat Piranha).
Cheers,
Raj Dutt
Voxel dot Net, Inc.
|