I'm really stuck with an issue regarding my service that's deployed through Envoy Gateway with the Gateway API. At first, everything works smoothly, but after some time, it starts throwing a timeout error when accessed via a browser:
upstream connect error or disconnect/reset before headers. reset reason: connection timeout
When I access the service from within the cluster using curl, it responds perfectly fine. So, it seems to only be an issue when accessed externally. This particular service is unique because it's the only one exposed on the apex domain. Here is the Gateway and HTTPRoute YAML configurations for reference. I'm confused as to why this happens especially since it starts out working perfectly. Has anyone run into something like this before?
1 Answer
It looks like you might need to dig into what’s actually failing here. Since accessing the service works fine from within the cluster, the issue seems to lie with the Envoy Gateway. You should definitely check the logs for any indications on why the connection fails externally. Timeouts can suggest that the service isn’t listening properly or that it's an issue with the proxy accessing it. Getting the logs will help clarify what's going wrong here.
The only log I found in Envoy during those times showed a 503 response with connection timeout details, which indicates that requests aren’t even reaching my app pod.