Add cht-user-managment-worker and redis to cht-user-management helm charts

[Hareet]

Resolves #19

A few things I need to fix before merging:

• Chart.yaml in cht-user-management requires version "0.2.0" for its dependencies. If we run helm commands with --dry-run --debug with "0.2.1" versions listed, the manifests dont get populated for cht-user-management and worker containers. Only redis is populated. Rolling that chart depedency back to version "0.2.0" populates all manifests.
• We have some nested dependencies (cht-user-management) uses single-base-app to create its templates, and there is some version mismatching that is happening.
• We need to fix this, because we want to release single-base-app "0.2.2" which would include options around if ports, services or healthprobes are needed and that would enable us to use single-base-app for cht-user-management-worker

Current workaround:

• Deploy helm charts with Chart.yaml "0.2.0" for all single-base-app
• After deployment is running, edit cht-user-management-worker deploy resource and delete the lines referencing ports.

Testing:

• cht-core 4.9 : hareet-test.dev.medicmobile.org
  • togo config uploaded, no seed data yet
• user man tool: hareet-test-users.dev.medicmobile.org

[mrjones-plip]

Thanks for reaching out for early feedback! Much appreciated.

While I'm still pretty (VERY!) rusty on how helm works, I parsed out some basics and had questions which I'm couching as "request changes" ;)

[mrjones-plip]

All the additions to this file seem bespoke to a dev install for Hareet in EKS. Is this correct? If yes, I'll do a deep dive on what looks wrong. If no, I'll wait 'til final revisions come in.

[Hareet]

Yeah, I've added some comments on what to change, and thought other environment variables were self-described. We can add more comments to clarify

[mrjones-plip]

It would be great if we could update the deploy steps to NOT depend on the deploy branch in CHT User Man repo:

Switch to the deploy branch and ensure the same values file from the prior step is in the values folder in deploy branch

Maybe that's for another PR though? Really - this is error prone and done manually for all prod pushes - a good recipe for a mistake to be made!

[Hareet]

So the deploy branch only contains the values yaml, and you want that moved to the main branch? Why is it in the deploy branch in the first place? Let's chat about it and put it in a different PR.

[Hareet]

medic/cht-user-management#213

Here's the values file we want to use for users-chis-tg deploy.

[Hareet]

@mrjones-plip Please approve.

It's ready for anyone to use. We'll scrub the default values that say hareet-test in another PR and fix our interpolation of REDIS_HOST, so it doesn't have to be manually entered.

Here are steps to use it:

• Git checkout repo, go into directory:
• helm dependency update
• helm dependency build
• helm -n <namespace> install <release-name> -f values.yaml .

If someone doesnt approve in time or Chart.lock interferes with above steps, and this needs to be rushed deploy:

• Git checkout specific branch cht-user-management-worker
• In Chart.yml (cht-user-management directory), uncomment the repository line and comment the other one out. For all dependencies.
• helm dependency update
• helm dependency build
• Change all the values to reflect your prod deploy
• install chart

@henokgetachew Is @paulpascal set up for Production EKS? We can have him deploy this.

[Hareet]

Tests are only failing because I require a version of base-single-app is also being merged in this PR. Then the dependencies can build and tests will pass, so lets ignore that for this moment.

[mrjones-plip]

@mrjones-plip Please approve.

Done!

It's ready for anyone to use.

That's awesome - thanks!

The measure of this being true is for anyone but SRE to successfully deploy by following the deploy docs - these seem to differ now from what you listed above.

@henokgetachew Is @paulpascal set up for Production EKS? We can have him deploy this.

nope! Ro too. It would be amazing to fix both their perms and see one of them can follow steps in this PR to deploy \o/

[Hareet]

@mrjones-plip Thanks. I'll test during my tomorrow morning if we can just run helm upgrade as per the docs, or whether we need the helm dependency steps beforehand.

[henokgetachew]

The following people already had access to the namespace users-chis-prod:
"fredmuiru",
"kennsippel",
"mrjones"
I have added Paul (User: paulpascale) and Ro (User: romuald) and pinged you in the PR - I have applied the changes so he now has access.

[mrjones-plip]

noting that Ro's access was fixed too