Hello, we’re currently testing a deployment of Play 2.7.3 using a Kubernetes cluster and Google Cloud Load Balancer. I’ve run into something strange that I’d like some feedback on. Our QA team kept running into random, but quite frequent 502 bad gateway errors, presumably a timeout between the load balancer and play instances. There was no error log or even any indication the request was received by the controller.
We suspected a keep-alive issue, but adjusting timeouts didn’t resolve the issue. The strange part was that the problem resolved itself after we tried switching the backend to netty.
I’m wondering if I’m missing something, or if anyone has had a similar experience?
Thanks,
Jonathan