Health check and Drain stop behavior



What is the expected behavior in regard to Drain Stop feature when health check is failing?

Assuming the following configuration:

Health check enabled (HTTP)
Drain stop time set to 1 hour
Active Cookie Persistence enabled (statefull application) 
2 servers (A and B), both healthy.

When a health check fails against server A while active sessions are still active, what is the expected behavior when users assigned to server A make further http requests?

A) Requests are gracefully forwarded to node A (up to drain stop time)
B) Persistence is reset and requests are forwarded to node B (without any delay)
C) None of the above (please add details)


1 comment

Mark Deegan

Hello Hugo,

Drain time will allow current open connections to maintain their transmission with the real server until the drain time expires. no new connections are allowed to the server while it is marked as down. Then after the drain time expires the real server is taken out of rotation and all open connections to this server will be severed.

See our document on drain time here