Does Health monitoring work without PROBE command in serverfarm submode?

[maranova]

Hi!

I'd like to know Probe command in serverfarm submode.

The Probe command is mandatory or optional?

I put 'probe 5000 tcp' in serverfarm submode and service demon in real server was up.

In 'show mod csm 6 reals', it said it is probe-failed.

Also in 'show mod csm 6 vserver' there are out of service on vserver.

To test my cofiguration, I remove 'probe 5000 tcp'.

After that, reals/vserver worked well.

Of course, i don't know how the server work,especially health check with L4 switch(CSM)

Why is it happen? It works well without 'Probe"?

Thank you.

Jung Hak Hyun

Seoul

[Gilles Dufour]

probes are used when you have multiple servers and you want the CSM to automatically detects the servers that are down to avoid forwarding traffic to them.

probes however are not necessary.

Here is a link to probes configuration

http://www.cisco.com/en/US/products/hw/switches/ps708/products_installation_and_configuration_guide_chapter09186a00801c5899.html#wp1024615

Regards,

Gilles.

[MARK BAKER]

Hi Jung Hak Hyun,

health probes are not manditory, but they do give you a better way of determining if your servers are truly healthy.

I noticed without probes being configured, when a server was shutdown, it was reported as failed in the CSM. I did some packet captures to see what the default health probe was. I knew the CSM had some way of detecting a downed server. I found that ARP is used by default when a probe is not configured and applied to the serverfarm. When the CSM doesn't receive ARP replies from the server, it lists it as failed.

If you have a web server with a stopped web service, as long as the server can respond to ARP requests, the CSM will continue to send user traffic to it. A properly configured HTTP probe would be able to detect this and take the server out of service.

Mark