This is a broad question about the Helm chart `dag...
# deployment-kubernetes
b
This is a broad question about the Helm chart
dagster-user-deployments
Here the deployments section is made up of array of deployments with names. The problem with arrays in Yaml is that they are replaced when layered with overrides, so if I have base values set in the base chart, and only want to update the tag of
deployment[0]
in a separate
values.yaml
file, it will completely overwrite my `deployment[0]`’th item
d
Hi Binoy, this is one option - you can do something like:
Copy code
--set-string dagster-user-deployments.deployments[0].image.tag=NEW_TAG_VALUE
when running the helm upgrade command, while also applying a values.yaml file to handle everything else other than the tags
as long as the square brackets are escaped properly
b
Sure but it makes it hard to track which user deployment is to be affected .. coding it to index based lookup vs name based lookups.
d
fair enough - maybe wrapping it a script that maps the names to indexes?
b
just out of curiosity, was this intentional or was the
values.yaml
file layering / yaml merging aspect missed. I am talking more in terms of using CI/CD to merge things in the code commit pipelines
modifying more parameters via cli
set-string
makes it harder to manage on the long run..
d
This is good feedback, I can file an issue for this
@Dagster Bot issue Making dagster-user-deployments a map instead of a list would make injecting tags during CI/CD easier
d
b
wonderful. Thanks @daniel