The current solution for a load balancer to determine the status of an AA web server is to build rules based on the AAF published API. Suggest adding a single healthcheck URL that performs some basic tests on the web server and responds "Success" or "Failed" (and the reason for the failure). Some checks that would be necessary would be status of the web server and status of an authentication request (perhaps configured through a test account in the admin ui). The goal is that if the URL returns "Failed", the load balancer would know to remove the web server from the pool.
by: Jared L. | over a year ago | Installation/Deployment
Comments
Fully agree. The current lightweight healthcheck of using "GET/api/v1/status" only checks that the webservice is runningbut not necessarily that all of the internals of that AAF application-layer on that server are healthy and able to service authentication requests. A smarter health-check URL would be one designed more like the one for NetIQ Access Manager that is informed by a battery of internal healthchecks (is a DB Master database reachable and operational, are at least one of the configured LDAP sources for each defined repo operational, etc).