[lvs-users] health checking on realservers

John Lash john.k.lash at intel.com
Mon Aug 9 18:29:17 BST 2010


I'm trying to check the availability of a service on a pool of direct-routed real-servers behind an LVS director. The service only listens on the VIP (not on any of the RIPs on the servers).

My problem is that since the real-servers only listen on the VIP attached to the loopback device, I have no way to address them to make sure that the service is usable from any location off of the real-servers themselves.

I've tested my configuration with another service, telnetd, that binds to all interfaces and there's no problem with the health checks and aside from this the cluster config works very well. I've set it up manually and with keepalived and as long as I use a server that binds to an externally addressable interface, it works just fine.

Is there a way to health-check this service within the LVS framework? I haven't been able to find any information about it out on the web. The best solution that I've thought of to this point is to use ssh (or similar) to run a command locally on each server to do a health-check.

--john




More information about the lvs-users mailing list