|
Hi,
Situation: LVS running 2 x 2 nodes on VMWare ESX. Dual NIC per LVS
box, one for normal traffic and one for heartbeat/sync-daemons
Now, i've come across some problems while running on VMWare because as
soon as one of the other virtual machines on the physical box starts
to use too many resources (mostly network), the LVS's miss
heartbeat(s) and a failover occurs. As it should, ofcourse. But this
is not a nice situation for us.
So while the vmware issue is being researched (we're going to have a
seperate vlan for heartbeats etc) i want to make sure a failover won't
occur again. So i was thinking about simply taking out the
inactive/slave node.
Is there something i need to be careful with? I already set
nice_fallback (or something similar) to 'off' but i can imagine there
are more steps i can take in order to make this as painless as
possible for the network users.
Thanks in advance,
Léon
|
