Helm chart persistent volume

currently looking at the latest version of the helm chart but it seems there is no built in way for the coder helm chart to have persistence so everytime the main coder pod dies everything with it i.e. users, groups, workspaces goes with it. any plan to add persistent volume support or keep settings permenant in the database?

Codercord•10/6/23, 7:37 PM

What are you creating this issue for?

bpmct•10/6/23, 7:39 PM

Hey! Coder is designed to run with a persistent Postgres database which holds all of this state. We should probably disable running Coder in Kubernetes with the "built in" database. We have some docs on running Coder with a persistent DB here: https://coder.com/docs/v2/latest/install/kubernetes

Kubernetes - Coder v2 Docs

Install Coder with Kubernetes via Helm

mavenOP•10/6/23, 7:43 PM

oh so when the pod starts up it uses a pod internal DB instead of the external DB i set up lol

mavenOP•10/6/23, 7:49 PM

nvm actually looking at the documentation looks like i missed a step

mavenOP•10/6/23, 7:52 PM

i created the secret with the postgres connection url, but failed to set the ENV var to the correct value.

mavenOP•10/6/23, 8:04 PM

this could potentially by a true false flag in the helm chart when external is true the pod errors out if the connection url isnt set. do you guys accept public PR's for review? would be interested in iterating over the helm chart

matifali•10/7/23, 8:19 PM

Yes sure. You can submit a PR.

Scott•10/8/23, 6:33 AM

I'd be interested in your solution, because I also just complained in another app's helm chart repo for a similar "it's up to you to config" kind of scenario like this one. Maybe I could take your solution for fixing their situation too.

(disclaimer: I've never worked on Helm chart internals before).

Helm chart persistent volume

Similar Threads

Similar Threads

Similar Threads