Index ¦ Archives ¦ Atom

Placement Container Playground 9

This is the ninth in a series about running the OpenStack placement service in a container. The previous update was Playground 8.

The container playground series introduced running placement in Kubernetes in Playground 4 and then extended it in Playground 5 to add a Horizontal Pod Autoscaler.

But it was very clumsy. For some other work, I've needed to learn about Helm. I was struggling to get traction so figured the best way to learn how things worked was to make a helm chart for placement. There already is one in openstack-helm but it is embedded in the openstack-helm ecosystem and not very playgroundy. So I set out to play.

The result of that work is in a pull request to placedock (since merged). A relatively simple helm-chart is built from the starting points provided by helm create. It started out simply deploying a placement service with an internal database, but through iteration it will now set up ingress handling and the aforementioned autoscaler with:

helm install --set ingress.enabled=true \
             --set replicaCount=0 \
             --name placement

(replicaCount=0 is used to signal "make me some autoscaling, not a fixed set of replicas".)

There's more info in the placedock README.

© Chris Dent. Built using Pelican. Theme by Giulio Fidente on github.