Hey, using Dagster for a while now, I’m loving it ...
# ask-community
p
Hey, using Dagster for a while now, I’m loving it daggy love But there is this one thing I fail to understand… We are deploying new repositories quite often, more than daily (for instance in test we deploy a separate repo per PR, to create an isolated staging environment). Unfortunately deploying a repo is still a part of the dagster deployment, not the user-code deployment. I feel like it’s missing the point of having a separate user-code deployment in the first place. What is your view on this? Are you planning to fully separate the user-code deployment in a future release?
1
Because in the end it is just the
workspace.yaml
that needs to be updated with the new servers, am I correct? The
dagster-dagit
and
dagster-daemon
could be set to automatically pull the workspace-yaml configmap whenever it gets updated. That would work right? But maybe I’m missing something 🤷
Actually mounted configmaps get auto reloaded by default except when using a configmap as a subpath volume mount. You appear to set the subPath here. I wonder if that is strictly necessary.
m
We are discussing literally the same thing today at work!
d
Hi Pieter - this looks like there's some overlap with the feature request here? https://github.com/dagster-io/dagster/issues/6295 (or at least that would be one way to accomplish this)
p
Very much related @daniel, thanks! Can you say something about when you are planning to pick this up? Rough indication maybe?
d
There's a bit more activity on https://github.com/dagster-io/dagster/issues/7139 which is also related, and I see an open PR for from a community member
p