rare nova
Hello, I am seeing an unusual issue with serverless loadbalancers. Basically I have a endpoint with one active worker, and a max of 3. When sending requests to the endpoint, the active worker does not appear to receive any of those requests.
Eventually, when enough of those requests have failed to resolve, the system scales out to a second worker, and that worker functions correctly (i.e. it receives requests). Not sure if this is related to the "active worker" or it's just a lottery as to whether the worker will function or not. Could be related to the issue @quick cosmos announced a fix for on the 22nd, but it's something I'm still seeing right now.
Any guidance on this would be greatly appreciated
uneven pagoda
spice mantle