What is not going well?

In a short word: Infrastructure.

Basically, Treehouse's community services have come about through experiments.

As a result of this, the infrastructure is entangled with my own personal infrastructure, although it is becoming less entangled over time due to the adoption of OpenStack on the hosts: Treehouse infrastructure now has its own dedicated kubernetes cluster, but further isolation work needs to be done before access can be given to more of our volunteers.

At present, the plan is to abstract most of these problems away with service accounts and IaC. Once the infrastructure is fully detangled, more "break glass" access will be given as needed to trusted volunteers.

The other problem is that we use kubernetes, and specifically the knative framework on top of kubernetes, with a custom traefik ingress setup. This setup is not particularly well-supported by upstream knative, although I am working on plans to upstream that work into knative directly, which should help a lot.

We were going to consider switching to Nomad, but Hashicorp decided to change the license, so we're probably not going to do that now.