I’m running Concourse v4.2.2 in Kubernetes, with Helm (Concourse chart v3.7.3).
I have deployed a second Concourse instance (let’s call it “blue”) alongside the previously running one (“green”), to perform smooth upgrade / switchover. I’ve realized that the blue instance is using the database from the green deployment. I am struggling to find out why.
When deploying with Helm, the configuration is passed via env vars. The database connection is configured with more than a few variables, it seems:
$ kubectl -n concourse exec -ti concourse-web-67fd4bf778-ljl4x env | grep POSTGR|cut -d’=’ -f1
I have noticed that CONCOURSE_POSTGRES_HOST is inconsistent with CONCOURSE_POSTGRESQL_SERVICE_HOST in the blue instance - the CONCOURSE_POSTGRESQL_SERVICE_HOST variable points to the database of the green instance. It is consistent in the green instance of course.
Grepping the code or googling for CONCOURSE_POSTGRESQL_SERVICE_HOST or parts of that name gets me nowhere (unlike with CONCOURSE_POSTGRES_HOST).
Any help appreciated, thanks!