Installing multiple helm charts in one go using simple bash utility

Ref: https://dev.to/aws-builders/installing-multiple-helm-charts-in-one-go-approach-1-5d1p

Ref: https://dev.to/aws-builders/installing-multiple-helm-charts-in-one-go-approach-2-using-helmfile-4e4h

In some of the air-gapped environments it is sometimes a bit difficult to use the tools/utilities available because moving things inside an air-gapped environment is a challenge.

Some of the environments are so secure that one may need to follow a whole process of getting all the security clearances and approval before using a tool/utility, which altogether is a nightmare.

I chose bash, the reason being it is pretty common among engineers and it is easily understandable.

The source code can be found here: https://github.com/sunnybhambhani/helmister

You can copy it, and tweak it based on your requirements.

A running Kubernetes cluster with proper permissions, here I have used minikube.

`kubectl` and `helm` installed on your machine to interact with the Kubernetes cluster.

A clone of helmister repository.

yq(required) and cowsay(optional) packages installed on your machine.

This is a small bash utility that can help to install and uninstall multiple helm charts in one go. The idea is inspired by `helmfile`.

Its usage is quite simple, just run the utility followed by the option like install or uninstall i.e. `./helmister [install/uninstall]`.

Under the hood, it calls `helm` binary. Therefore it is kind of a wrapper around `helm`.

It consumes a `config.yaml` file which contains all the necessary details about the releases, and common parameters.

This supports both `oci://` and `https://` helm registries.

PS: In the future, I am planning to add some more options/functionalities to this.

helmister, this is the script written in bash you can just cat and see what all things it contains.

config.yaml, this is the main configuration file or you can call the state file which contains the list of all releases you want to install in a cluster, plus it also contains some additional key:value pairs that are generic and common across all the releases. I first kept this configuration file in csv format, but later decided to convert it to yaml because it is more readable.

logs, this is a directory that holds the logs of this utility, it contains the information about the execution of the last iteration, plus any archived/past logs (if required for reference).

README.md, contains bit of a documentation about this utility, and what options are present.

values, this is the directory where all the values files are placed (it can be anywhere in your system but for simplicity, I have placed them in the same directory).

The initial key:value pairs are common across all the releases.

charts, is an array that contains a list of what all releases need to be installed in a Kubernetes cluster.

dry_run, it is a boolean [true or false], and if true, none of the Kubernetes resources will be created it will just do a dry_run.

create_namespace, it is a boolean [true or false], and if true it will automatically create a namespace for the release specified in the charts array.

wait, it is a boolean [true or false], and if true it acts similar to `helm --wait` wherein the shell will be kept occupied until all the Kubernetes resources are created.

timeout, it is a boolean [true or false], and if true it acts similar to `helm --timeout=20m` i.e. if all the resources are not created within 20 mins the execution will fail. By default, I have kept the timeout as 20 minutes which is more than enough.

charts, it is an array that contains details around individual releases. Except for version and namespace all the key:value pairs are mandatory.

release_name, this is the name of the release.

chart_name, this is the name of the chart that needs to be installed.

chart_repo, this is the helm registry where the chart is located. It can be any `oci://` or `https://` registry.

values_file, which contains the path of the values file for individual releases.

version, this is the version of the chart that needs to be installed. This is optional and if not provided, It will consider the `latest` chart version.

namespace, this is where the chart will be installed. This is optional as well and if not provided it will be installed in the `default` namespace.

I will use the same `config.yaml` which will install one helm release from an `oci://` registry in `default` namespace (since I haven't specified any namespace for that release) and another one from `https://` registry in `argo-cd` namespace. Note the version as well for nginx, there as well I haven't specified any version, it will pick the latest available one automatically.

This is my cluster's current status.

Actual stdout output can be found here: https://github.com/sunnybhambhani/helmister/blob/main/artifacts/sample_logs/during_installation_on_stdout.log

Here you will see that it provides all the minute details of what exactly it is doing.

For example: Generic values

Here I have marked create_namespace as true because I don't already have argo-cd namespace wherein I want to install argocd release.

Next, you will see, chart/release specific values of all the items in charts array one by one:

Here I have explicitly added an additional check that will check if the pods are up and healthy (this is just specific to pods it won't consider any other k8s objects).

It will continuously check for next 20mins and will check every 5 secs. If the pods are still in non-running state it will terminate the process.

Once all the pods are in running state and it has deployed all the releases it will show a successful message.

Once this is done, if required you can see the logs as well from the logs directory. A sample can be found here: https://github.com/sunnybhambhani/helmister/blob/main/artifacts/sample_logs/actual_log_file.log

And here is my cluster's current status now:

If you want to uninstall it, simply do:

It will get everything cleaned for all the releases that are specified in `config.yaml`.

The sample log can be found here: https://github.com/sunnybhambhani/helmister/blob/main/artifacts/sample_logs/during_uninstallation_on_stdout.log