Feature: Add a Helm Chart for Vigilo #21
Conversation
We will be using a Kubernetes CronJob to run Vigilo periodically. This add the initial template for the CronJob.
This will template out the name CronJob name and the container image configuration. Vigilo is not currently publishing images to Docker Hub, so there is no default image to use.
To run Vigilo, we need to have a `url.json` file and a `lh-config.js` file within the container. This can be seen in `dev-scripts/run-docker.sh` file when running the docker container directly. Therefore, in order to add these files to the cronjob container we will volume mount them. To do that, we need to create a configmap for each of these files to be mounted. As such, this adds a configmap template for the `lh-config.js` file and `urls.json` file which can then be modified in the `values.yaml` file.
The last bit to getting the helm chart working is to add a secret for the DataDog credentials (i.e. the .env file). This adds a secret template and updates the cronjob to include the secret. The creation of the secret can be turned off and we can point to a different secret if needed.
ardelato
added
the
external_block
There was a problem hiding this comment.
Are there any helm linters that we can enable in this repo?
maybe https://github.com/marketplace/actions/helm-chart-testing?
There are a couple things I want to add to help automate some of the Helm stuff. For instance, the usage of That said, I think we should create follow-up issues to address the testing aspect. |
This is still in the early phases for us, so manual testing will be the initial validation to the charts. That said, one thing I should point out is that unlike some of the other Helm Charts we've interacted with, this uses a CronJob instead of a |
|
QA 👍 I was able to get Vigilo integrated and working in the test cluster without any modifications to the current Helm Chart.
|
Description
This mainly follows examples I've seen from other Helm Charts. The entire premise to be able to routinely kick off a lighthouse audit as part of our Kubernetes cluster. This will detach the dependency of having Lighthouse running on our Ubreakit machine and instead will be ran through a Pod.
As such, it will be a lot easier if we have a Helm Chart for all this instead of manually creating the manifests ourselves in our FluxCD repo.
QA
Note
This will be difficult to QA as is, we will need to manually test by integrating the Helm Chart into our current Test cluster.
GitRepositoryFluxCD resourceCloses: #16
Connects: https://github.com/iFixit/ifixit/issues/49455