Hi,
I've got a proxy connecting to two 'servers' that are connecting from the
HAProxy to the remote servers via an ssh tunnel.
Here's an example:
listen portal 127.0.0.1:8085
mode tcp
balance source
timeout check 3000
server i-e4014f9c localhost:58919 check
server i-76a4ea0f localhost:38559 check
localhost:58919 and localhost:38559 are forwarded to separate remote servers
via an SSH tunnel.
What I'm seeing is that I can take down the remote service and HAProxy doesn't
detect it, but if I take down the ssh tunnel it detects it. I'm wondering if
there is some configuration that I can do to have HAProxy detect when the
remote server's service goes down?
I've seen this possibility of using a separate check port:
http://blog.christory.net/2009/09/15/building-an-easy-and-scalable-load-balanced-high-availability-web-hosting-solution-part-one-the-front/
but I'd prefer to not do that because it just adds further complexity to the
setup (more ssh tunnels).
Thanks,
Alex