A quick post today, about how the A10 load balancer handles the recovery of a server based on health checks. Kind of a warning actually.

A10 Health Monitor Configuration

In order to determine when a service should be considered “up” again after a health monitor has marked it down, it requires that the service respond correctly Consec times in a row, where Consec represents the “Consec Pass Req’d” value in the Health Monitor configuration, as highlighted below:

By default, ACOS (the A10 OS) sets Consec to 1. In other words, a single successful health check would bring the server back into rotation. I have no idea why A10 think this is a good value to choose, but I share it here as a warning because this is really not a value I would ever recommend anybody to use in production. If you have a service that is bouncing up and down, the last think you want is for it to be marked up and down repeatedly.

Light At The End Of The Tunnel

The good news is A10 do allow you to change this global default setting. The settings are within Config > Health Monitor > Global, so I would recommend changing it right after installation.

BUT – A WARNING!

If you have a live system, do not change the global value unless you understand how it will affect the existing configuration of health monitors. What do I mean? I’ll explain in another post, but be warned, you probably want to wait.

30 Blogs in 30 Days

This post is part of my participation in Etherealmind’s 30 Blogs in 30 Days challenge.