feat(cogstack-cohorter): Add helm chart for cogstack-cohorter

[jocelyneholdbrook]

No description provided.

[alhendrickson]

Approved if you want to merge it and publish it out.

I think the main comment would be if you can redo some based on the output of helm create .., which should generate a bunch of other stuff like the test, httproute etc. Convention seems to be keep it all super standardised

[alhendrickson]

Move under the webapp?

[alhendrickson]

Well - I'd approve if it passed the check anyway!

------------------------------------------------------------------------------------------------------------------------
==> Logs of container cogstack-cohorter-helm-z581pqgk5y-webapp-58587cd997-fcndf
------------------------------------------------------------------------------------------------------------------------
[webapp] ERROR: No data found at /usr/src/app/server/data.
[webapp] Mount a directory containing snomed_terms.json (and related files)
[webapp] or snomed_terms_data.tar.gz via a Docker volume:
[webapp]   -v /your/data:/usr/src/app/server/data

[alhendrickson]

You can add a file like your-chart/ci/ci-values.yaml to set a value for the github check to use

I'm, thinking its this one?

  env:
    RANDOM_DATA: "false"

I think the file suffix must be *-values.yaml, and you can put multiple files in if you have different scenarios to test.

[alhendrickson]

Hey I think this is needed for startup right?

Can you make this the default then?

The aim would be to be able to helm instal oci://...cohorter and it bring up the app with no values or prerequisites required at all.

[alhendrickson]

For a later improvement, potentially just chang the actual code, eg put this script into the entrypoint or just change the code itself to handle it. I should also be able to docker run .my-image with nothing else required for it to startup successfully

[jocelyneholdbrook]

thing is I don't want to start supplying the SNOMED data, at least not at this stage. This is how the app behaves today i.e. you the user provide the SNOMED data even when you ask it to generate random patient data. Granted, I think it should operate in one of two modes - fully customised with all user-input data or all generated with it's own code mappings (this could be SNOMED or something made up).

[alhendrickson]

I'm fine with this for now to merge in

Looking ahead - I'm looking for all the apps to all start up with sensible defaults, so nothing has any requirements (which is a big change from today in general...). This seems to be the pattern that I'd want to follow, for example I can run nginx but it is literally useless, it shows its default index html and 200 OK, it doesnt say "give me a conf file"

So here, even you want to avoid the data generation, I'd still see if we can have the container status "running", even if you then open the page and it says "200 OK, but insert data"

[alhendrickson]

I think you can remove this now it is in ct.yaml

[jocelyneholdbrook]

removed

[alhendrickson]

Approved ! Cheers for the changes. I think anything else mentioned could be done later/never

[alhendrickson]

Can we default to a smaller one? So it can demo on small hardware / CI easily - for example the Azure tutorial is bringing up just 4GB nodes - would prefer it running with bad results to then justify additional recommended hardware

[jocelyneholdbrook]

yeah - happy to do this as part of a separate task. Would need to assess what can provide comparable results with less etc. At a glance, I'm thinking medgemma (I've heard David mention this) or qwen 3. thoughts?