CLUSTER - Status page who check the working state of LLNG
Summary
The idea is to tell Keepalived service that LLNG not working fine. (EX: memory issue, or mongodb issue have generate an error 500) but nginx not fail-back even if there something wrong... The idea is to add a HTTP_GET healthcheck to told keepalived service to force a fail-over on the backup-node. That easy to do.
But to get it working on LLNG side we need a status page who will try to authenticate an (defined test user) and return a result like: Everthing seem to work! if not, another message. It's more like an unit test page who call on demand (every 30 seconde by keepalived service)
Design proposition
auth.exemple.com/check_state
return a simple HTML page with the result. Note: the result should not change between version to avoid failover when upgrade to a new version.