Hello, following this https://docs.dagster.io/deployment/guides/kubernetes/deploying-with-helm#configure-your-user-deployment but not sure how to add the environment variables. I have a local .env file which needs to be available there. Anyone else faced this?

Hey abhishek - somebody from the team will get to your question later today. Most of the team is active during US business hours on weekdays so we appreciate giving it a day before pinging the thread

the user code helm chart provides three different ways to pass env vars to your deployment: https://github.com/dagster-io/dagster/blob/master/helm/dagster/charts/dagster-user-deployments/values.yaml#L73-L100

Here's a command that will create a kubernetes secret for you from a .env file:

kubectl create secret generic yoursecretnamegoeshere --from-env-file=.env

Then you can reference that secret in the helm chart using the envSecrets field in Adam's link

To use it in the code, I can still use os.getenv()?

yeah, it still includes it as an environment variable

Thanks. I will try this.

Ah is there a DAGSTER_HOME in your .env file? Try taking it out

if I change something in the values.yaml, should I just run the below command to refresh the pods?

helm upgrade --install dagster dagster/dagster -f values.yaml

I actually added

--inject-env-vars-from-instance

argument here as shown in the snapshot below but now am trying to remove it. However, my changes in values.yaml aren't propagating.. I keep seeing the pods in status

CrashLoopBackOff

and as soon as I delete them, they restart but I don't think my changes in values.yaml are being taken. For the above issue, I just added a single secret instead of using my .env file and my .env file doesn't have DAGSTER_HOME. The DAGSTER_HOME error is coming in one dagster-dagster-user-deployment- pod and the other has some other code related error, if that helps.

Yeah, if you change the values.yaml then a helm upgrade should apply it

this is the first time I'm working with k8s and helm. Learning as I go.. what would be the command for uninstall?

helm uninstall: https://helm.sh/docs/helm/helm_uninstall/

I shouldn't have to delete the pods, right? helm upgrade should just update it, right?

That's corect, helm upgrade should be updating the configuration of the pods

okay, removing the dagster release and then adding it again helped. back to the code related error. I saw that we are not able to use Upper Case environment variables names in the values.yaml file, is this correct?

I've used upper case environment variables before without issue

All good now! 🙂

that's right, you can have multiple user code deployments. you can also add additional jobs to the same user code deployment though if you want

Got it. I was talking in terms of opening this up to other teams, DS/ML/DE.. now different teams can just add their deployments. From CI/CD POV, we would need to run that helm upgrade command to refresh the server each time docker is updated? What are the best practices for CI/CD here?

I think that's maybe worth a new post

couldn't agree more, thanks so much Daniel!!