Helm values schema gen Plugin
tldr
$ helm plugin install https://github.com/karuppiah7890/helm-schema-gen
$ helm schema-gen <values-yaml-file>
GitHub repo for the plugin - https://github.com/karuppiah7890/helm-schema-gen
Longer version
Today I finally created a Helm plugin that I wanted to for so long. The repo is here
In this post I’m going to tell the story behind the plugin. First the what, then the why, and a bit of how.
The What
Recently, helm released their version 3 which was a big thing. It has some really cool features and you can expect more in the v3 series
One of the features in v3 is, checking the schema of values.yaml
using json schema. The schema is present
inside the helm chart as a file called values.schema.json which
contains the json schema. So, now if you make mistakes in the
values.yaml you will know it if you have the schema file with proper
schema definition. Schema validation check happens whenever you run
lint, template, install or upgrade. You can read more about it
here
Now that you know about this feature, I can tell about the helm plugin
I created. As you can already see in the title, I created a plugin to
generate a values schema json for a given values.yaml.
The Why
The why might look very obvious. But I actually have multiple whys and not all of them are obvious I think. The first reason for creating the plugin is - ease of use. So, when I saw this feature in Helm v3, I was quite excited to check it out. I wanted to try it out with the default chart that comes with this command:
$ helm create test
But when I saw the big values.yaml in the test chart, I was
confused as to how I would write a very big schema file for it.
I was also thinking why the default chart did not have a schema
file too, so that I could try it out. I
asked the maintainers about this
and they felt that it’s better to keep the advanced features out of the
default helm chart, so that it doesn’t confuse the first time users.
So, I set out to try the feature by writing the schema file by hand
and by checking the docs on json schema
and the
example in the helm docs.
I didn’t write the whole schema file for the given values.yaml. I had
just written the schema for some keys and checked how it works while
linting and templating.
And then suddenly I had this idea! Why do I have to write this big
thing by hand? It’s so tedious and not easy. Also, I have the
values.yaml, why can’t I detect the type of each and every key and
value and create the json schema? And that’s how it all started.
Also, there are other reasons to create the plugin - many might still be using helm v2, and it will take some time to move to helm v3. But the migration to helm v3 should be really easy. That’s why the maintainers created helm-2to3 plugin for the easy migration. People migrate to new versions for the features and support. If you don’t know, Helm v2 support will stop at some point. And Helm v3 has really good features like values schema json, support for pushing charts to OCI (https://www.opencontainers.org) registries, removal of tiller and some more!. Now, some of these features might need some work before you use the feature. This the amount of work needed to adopt a feature a sofware offers. In the case of the values schema json, this is the work where you sit down and write the whole schema by hand, which will take a lot of time if you maintain a lot of charts. Also, why do by hand when a computer can do it for you? 😉 So yeah, that’s another motivation and the why for creating the plugin. Easy adoption of new features. It’s pretty important - because, marketing the new version of a software is one thing, which leads to people seeing it. And when more people see it, it means good marketing has been done. And then when people use the new version of the software, then that’s conversion, and that’s what matters. And that happens only when adoption of the new version of the software is easy. Adoption includes being able to migrate to the new version easily and also being able to use the new features easily, which is usually the reason to even upgrade, apart from support. So, this plugin will help people move to Helm v3 and be able to use one new feature easily - the value schema json check feature.
The How
I’ll tell a bit of the how I created the plugin.
When I got the idea, I immediately started checking out golang
libraries that can help me with finding the type of an unstructured
object (values.yaml data). I
found one good library,
but it worked for structured data like many others. I was
meddling with it to make it work for unstructured data. And then I left
it at some point. And then after a month, I recently started working on
it. I finally made the library work
in my fork.
And then created a separate repo for the command line tool for the helm
plugin. Now I spent hours checking how it’s all done by checking code
of other helm plugins. And I had decided to use urfave/cli
at first, but the UI for it seemed a bit different, compared to the
natural UI of cobra which I saw in
other plugins. It felt like I had to customize quite a bit in
urfave/cli library, so I just stopped and switched to cobra.
And then I chose to build the simple binary for the plugin. I ran it
a few times and it worked. And then I wanted to create a release
strategy, so that people can install the plugin and use it. I used
goreleaser to release the binary, and
for CI/CD I used GitHub Actions.
I had to create a shell script for installing the plugin. I noticed
many other plugins using a script that they had written or copied from
another plugin. I didn’t want to copy it or maintain it really, and I
knew godownloader is
cool, so I just used it and wrote a two liner script on top of it for
customization. And then I released it finally!
I tried out the plugin and everything worked really smooth! And then I decided to write a small blog post about it, and it turned out to be this big 👆! 😛
That’s all! Have fun trying out the plugin!
$ helm plugin install https://github.com/karuppiah7890/helm-schema-gen
$ helm schema-gen <values-yaml-file>