helm chart - overriding default values for router

[dan-catalano-vc]

I am reluctant to open a ticket, as I think this might be best for a discussion. Mainly because I might be missing something that was done intentionally.

I've been working through the Deploy to Kubernetes documentation as well as the Configuring the Apollo Router documentation. I noticed the docs under Command-line options said, "Where indicated, some of these options can also be provided via an environment variable. If an option is provided both ways, the command-line value takes precedence." This makes sense and follows standard practice.

The problem being experience is that there is no way to override passing in the --config argument (or for that matter --hot-reload), as one is defined in the default values.yaml file and the other in the deployment.yaml template. This is because .Values.router.configuration is always populated as a result of the default values.yaml file. Furthermore, it appears there is no way to completely override (set to null) the .Values.router or .Values.router.configuration values without causing a Helm error (or having no effect).

Onto why this is an issue. We could definitely put all of our configuration inline under .Values.router.configuration, I believe it'd work. But the intent is to use multiple layers of values.yaml files for our different environments' configuration and as such, it'd be best if could use multiple values.yaml without having to completely duplicate the router config. Lastly, if we didn't want to use --hot-reload, it appears there is no way to remove that argument.

Thank you for your time. Happy to contribute to the project to fix this issue if it appears to be a defect. Cheers.

[dan-catalano-vc]

cross listing issue #4781

[garypen]

Hi @dan-catalano-vc Thanks for starting this discussion.

aside: I wasn't a fan of making the router run in --hot-reload mode by default in our helm chart, but I was out-voted by other members of the community on that one.

The main problem with making the changes that you propose is that they are incompatible. This would require us to make a major revision of the helm chart so that existing users (who rely on, for instance, --hot-reload behaviour) aren't surprised.

The good news is that I think you can achieve your goals without requiring changes to the router helm chart.

We use the router chart as a subchart and we have multiple values.yaml files for our different (dev, staging, prod) environments. We apply those on a per environment basis to achieve the "override" behaviour you are looking for.

To remove --hot-reload, we do this:

router:
  <blah....>
  router:
    args: []
    configuration:
      <blah...>

In the above, we have a chart (in-house-router), which has various dev,staging,prod yaml files and the sub-chart configuration is under the first router:. The second router: (which may be confusing) is the router: section from the main router helm chart.

Here's an example command that we run to template out the chart and make sure we are seeing the right thing:

helm template --dry-run --debug helm/chart/in-house-router  -f helm/chart/in-house-router/dev.yaml

You can see we are just specifying our dev.yaml file which contains all of our dev "overrides" to the values in the router helm chart.

Is that helpful?

[garypen]

I'm closing this since there's been no activity for almost 3 months.