KongIngress: Unexplained delay applying ingress rule / plugins


#1

I am just wondering about whether I am witnessing a bug, or behavior by design.

I have Kong Ingress controller running, cleanly no rules. I have a kubernetes yaml files with my ingress spec, ordered correctly. It applies KongIngress, KongConsumer, KongCredential, KongPlugin, and finally an Ingress.

In some cases I do kubectl create on that file and Ingress controller processes… I then look at the postgres database tables for Kong to see if the configuration applied correctly.

One case I see the routes, upstreams created, but the Kong Plugins they were not showing in Postgres plugins database table. This means our Key Auth plugin was not applied and it was just routing traffic without key requirement for a few minutes. Then a few minutes later the KongPlugin appeared as well. I actually found it through this case when I did a test-post without a key and it did not return 401 unauthorized.

Other case on deleting the Kubernetes resource definitions, I could see some stuff being deleted but a route remaining. After a few minutes that would disappear as well.

I am not sure if there is a designed latency between application of Kubernetes resource and Kong writing it to the database. The Ingress controller logs looked to be OK, no errors that stood out. Just a long delay for things to be applied and not all things applied at the same time.

Using Azure AKS, 1.11.3 Kubernetes, Ingress Controller 0.2, Kong 0.14.1