I have a weird bug setting up tyk in kubernetes. When trying to curl tyk-dashboard endpoints (for example http://:3000$URL/admin/organisations ) about 40% of the time the endpoint reponds correctly the other 60 I get Connection refused. I can fire off the requests as fast or as slow as I like and it doesn’t seem to make a difference. The dashboard container appears to be running correctly and no restarts are logged.
Anyone seen anything like this?
Should note this is running in minikube so may also be a minikube issue
Could you confirm how many dashboard instances you have up and running? Are they being exposed via a service?
Are you able to share your k8s manifest for the dash & service?
I managed to solve this, it was misconfiguration causing the service to look for three endpoints, only one of which had a working version of Tyk, so it was round robining to two non exsistent instances. Whoops