Forum Discussion
michael_trott_1
Nimbostratus
Nimbostratusautomatic postponement of health monitoring
hi all,
We have a pool that has a specific health monitor applied, we have the issue that when the host performs certain batch tasks, it fails to respond to the health checks quickly enough, F5 decides the host is down and masks it accordingly. All load is then sent to the second host, & with the sudden jump in load kills the seconday.
We could change the timeout of the health monitor, but this then fails to mark the node as being down quickly enough, when it is actually down.
Wondering if there is anyway to notify the F5 with a control message from the host in question to momentarity suspend the health checks, & when the batch file is complete then turn monitoring back on?
thanks
1 Reply
- Hi Michael,
I'm not trying to be rude, but that sounds like a problem, but not a BIG-IP problem. If your secondary server can't handle the load, what happens if you lose a server? You'll be down hard.
iControl will let you do what you're looking to do - you can set the time between checks to a long time and then set it back utilizing the LocalLB::Monitor interface, but that doesn't solve your underlying problem - that you risk losing all connectivity to the pool if a single server goes down. I'd look to upgrade or add another server, personally. That way you won't get called in the middle of the night if one server drops off, and you won't lose business because customers could not get through.
Of course, I don't know your environment and it's entirely possible that you aren't in a position to add or upgrade, but if you can, I would.
Don.
