AlphaLB has an optional health check method. The health check 'pings' your application and checks for a valid response within 10 seconds (not adjustable in the configuration file). If no response is received within that time frame, the load balancer will demote that Alpha Instance by delivering less new traffic to that Instance. Users currently on that Instance are not disrupted. If the Instance continues to under perform, the load balancer will stop sending new traffic to that Instance altogether. It will, however, continue to ping that Instance and when performance has been restore, the load balancer will start delivering new traffic to that Instance.


This method works very well for normal Alpha applications. But if your application is performing poorly overall and all Alpha Instances are taking longer than the 10 second maximum, AND you are willing to allow your users to suffer a delay of over 10 seconds, then you should disable the health check feature because eventually the health check monitor will shut down all of your Instances.


The lines in the configuration file that initiate health check are shown below. You can define the path as simply "/" which means the health check module will ping your default page, typically index.a5w. But it is a better option to create a special A5W page in your application just for this use. Each and every hit (ping) to whatever page you select will be reported in your Alpha Access log, and that can add up. Creating a special page will allow you to purge those many records from your logs and will not create a false impression for your analytics.


Comment out (#) each of the lines below if you want to disable the health check feature.


[backends.backend1.healthcheck]

path = "/"

#path = "/optional_healthcheck_page.a5w"

interval = "30s"