I have a problem with recognising a service as "Alive" rather than "Dead" on first keepalive attempt after a reboot.
The root problem appears to be the fact that the first keepalive that is executed from the "Service" is attempted before the physical port has become active.
This means that the first keepalive fails and the "Service" is left in its initial "Dead" status until the next keepalive is triggered at the keepalive frequency.
We have a requirement that if a service is noticed to be "Dead" it is taken out of Load Balancing algorithms.....and stays out!
The only way we have acheived this so far is to run a script under cmd-sched every minute that greps the output from a show service command for "State:" and tests the next field for "Dead". If it finds a Dead status, the script then removes the service from the rule (after adding a failover service).
*** First question: Is there a failover method that will take a service out of content rule and leave it out?
In the meantime, our method works fine....until....the system is rebooted (perhaps without operator intervention as in a power cycle).
When the system is rebooted the services all start in "Dead" status. Thus if our script runs before the first successful keepalive the service is taken out (prematurely) and manual intervention is then required to re-insert the service if it is really "alive". We have alerts for service transitions and our "operations" people can do this if required but it would be best to avoid it if possible.
One method I have been trying to configure is to use a "first time after boot" flag to insert a pause in either the keepalive scripts or the cmd-sched scripts. I can't find one!
*** Second Question: Is there a method from a script to determine if it is the first time run since boot.
*** Third Question: Is there a method to set a variable at boot. (and reset it via script)
*** Fourth question: Is there a method to carry a variable (which would not exist after a reboot) across iterations of a script run from cmd-sched?
