Ming-Ching Tiew wrote:
I noticed that openvpn client ( subject to validity of my observation )
will handle failover in round robin fashion, if server1 fails, it goes to
server2. If server2 fails, it goes down to server3 and so on until
it reaches server4 before it will try server1 again.
But the fact is that all these servers do not have equal bandwidth
and do not have equal connectivity quality. As a matter of fact,
they are in decreasing quality/bandwidth. Therefore when server1
goes down and come back up again, I would prefer the openvpn
client to go switch from server2 back to use server1 again,
irrespective of whether server2 is still available !!!
Yes -- and my last answer was in fact responsive to your question.
Tools are available for managing services across multiple servers such
as to shut down all but one copy of the service. If server1 is the only
one with a copy of the OpenVPN daemon running, then all the clients will
fail over to it.
UCARP (which expects your servers to be on the same subnet, and is thus
not ideal for your situation) is onesuch tool, and the one that I use.
Others such as failoverd and heartbeat are available. If you find one
which is ideal for your configuration, please report back.
OpenVPN mailing lists