On Dec 29, 2015, at 3:52 PM, Mark Moseley <moseleymark@gmail.com> wrote:
Since the home_server is by necessity set to 'tcp', radiusd errors out with a message that 'status_check' must be set to 'none', if you try to set it to anything.
Because TCP is a reliable protocol. You don't need to do status checks. If the connection is up, the home server is alive.
Does that mean it's not possible to do failover with radsec proxies?
It should be possible to do fail-over. If all of the TCP connections are down, the home server is down.
I've got a home_server pool set up with two nodes. When I take down the first-listed home_server with tcpdump running, the freeradius on the 'client' side just keeps trying to hit the now-disabled server and never attempts the secondary. Presumably I'm missing something but google is no help so far.
It should do fail-over when it determines that the first home server is down. Do you have a simple configuration which reproduces this? And... as always... the debug output? Alan DeKok.