ACE 4710 - Prevent Failback to primary server when probes fail

[malmgren]

I want to configure my ACE so that if a probe fails, it fails over to the backup rserver, BUT it won't failback to the primary rserver until manual intervention is complete.   The problem is we don't want an rserver to fail and failover to secondary, then failback to primary, repeat... (flip-flopping).   I want to be able to have time to get on the server and find out what may have caused the probes to fail before it fails back.

Is this possible?  I looked through the LB configuration guide and didn't see anything that might accomplish this.  What is the CLI to accomplish this?

We're running version A3(2.4.

Thanks!

[stmccabe]

Hello,

You have the option to configure the sticky option for the backup serverfarm.  Any existing client connections and new connections that have an existing sticky entry in the stick db to continue to connect to the backup serverfarm, however any client connection that doesn't have a sticky conn or existing flow will get LB'd back to the primary serverfarm when it becomes available again.

Reference:

http://www.cisco.com/en/US/docs/app_ntwk_services/data_center_app_services/ace_appliances/vA1_7_/configuration/slb/guide/classlb.html#wpxref39283

[malmgren]

Hi, I understand the sticky configuration, but the problem with that config is the primary rserver (the one that failed) is then operational again for new clients.  My question was - how do I keep it "down" until I re-activate it?    I don't want any clients (new or existing) going to that server until we had a chance to analyze the problem and/or correct it, then manually bring it back to operational state.

Thanks, Matt