Forum Discussion
NimbostratusAutomatic node failover in Pool
We support a client which has four (4) nodes in a pool for which they only ever need two (2) nodes enabled at any given time, and the pool needs to maintain high availabilty for thier customers.
What methods are available to be able to automatically enable and dissable the two sets of nodes on the one pool during pre-defined maintenance windows and which can be controlled from the developer/server side?
I was thinking a custom monitor with some descision logic (including high availability and date/time considerations) to enable and disable the pairs as required?
This particular setup is a very basic LTM configuration which has one Virtual Server, one Server Pool, the four (4) nodes mentioned as well as some redirects rules in a policy and no iRules
Kevin_Davies_40Nacreous
There is a built-in function on the HTTP monitor called the Recieve Disable String. If the string appears in the healthcheck response from the server, it will disable itself in the pool. This happens automatically as soon as the monitor sees the string. If you use this you can control which servers are available from the server end.
This disable does not kill traffic. It allows existing and persistent connections to continue until the persistence records expire. The server will appear with a black circle as the status on the F5 when this happens. It is identical to an F5 admin disabling the server in the pool.
You must have a normal working healthcheck before you add the Receive Disable String. So for two servers you don't want to be up, just add the disable string to their healthcheck page, or switch to a healthcheck page which has the normal receive string and the disable string. This will make them unavailable in the pool.