* define Ntfy ingress (naive implementation) based on current target
* use patched Ntfy Helm Chart
* create Ntfy main user only if needed
* add info logs
* better error bubbling
* instrument feature installations
* upgrade prometheus alerting charts if already installed
* harmony_composer params to control deployment `target` and `profile`
Co-authored-by: Ian Letourneau <letourneau.ian@gmail.com>
Co-authored-by: Jean-Gabriel Gill-Couture <jg@nationtech.io>
Reviewed-on: https://git.nationtech.io/NationTech/harmony/pulls/107
The multiprogress wasn't used properly and leading to conflicting progress bars (within our own progress bars, as well as the log wrapper).
This PR introduce a layer on top of `indicatif::MultiProgress` to properly handle sections of progress bars, where we can dynamically add/update/remove progress bars from any sections.
We can see in the demo that new sections + progress bars are added on the fly and that extra logs (e.g. info logs) are appended on top of the progress bars.
Progress are also grouped together based on their parent score.
Co-authored-by: Ian Letourneau <letourneau.ian@gmail.com>
Co-authored-by: johnride <jg@nationtech.io>
Reviewed-on: https://git.nationtech.io/NationTech/harmony/pulls/101
The CI pipeline (`./check.sh`) was failing because of test errors, which was caused by the instrumentation framework complaining that no subscribers/listeners were registered.
Instead of setting up all tests to run with a dummy subscriber, move the implementation of the instrumentation behind a feature flag so that it runs only for tests.
There's a catch though: the `#[cfg(test)]` directive works only when directly testing the crate. If a crate `A` depends on another crate `B`, `B` will be compiled as usual (aka not in test mode) which will not trigger the `test` flag.
So we need to introduce our own `testing` feature flag for `harmony` core and import it with that flag (only during dev/test).
More info: https://github.com/rust-lang/rust/issues/59168
Co-authored-by: Ian Letourneau <letourneau.ian@gmail.com>
Reviewed-on: https://git.nationtech.io/NationTech/harmony/pulls/102
First step in a direction to better orchestrate the core flow, even though it feels weird to move this logic into the `Score`. We'll refactor this as soon as we have a better solution.
Co-authored-by: Ian Letourneau <letourneau.ian@gmail.com>
Reviewed-on: https://git.nationtech.io/NationTech/harmony/pulls/100
Introduce a way to instrument what happens within Harmony and around Harmony (e.g. in the CLI or in Composer).
The goal is to provide visual feedback to the end users and inform them of the progress of their tasks (e.g. deployment) as clearly as possible. It is important to also let them know of the outcome of their tasks (what was created, where to access stuff, etc.).
<img src="https://media.discordapp.net/attachments/1295353830300713062/1400289618636574741/demo.gif?ex=688c18d5&is=688ac755&hm=2c70884aacb08f7bd15cbb65a7562a174846906718aa15294bbb238e64febbce&=" />
## Changes
### Instrumentation architecture
Extensibility and ease of use is key here, while preserving type safety as much as possible.
The proposed API is quite simple:
```rs
// Emit an event
instrumentation::instrument(
HarmonyEvent::TopologyPrepared {
topology: "k8s-anywhere",
outcome: Outcome::success("yay")
}
);
// Consume events
instrumentation::subscribe("Harmony CLI Logger", async |event| {
match event {
HarmonyEvent::TopologyPrepared { name, outcome } => todo!(),
}
});
```
#### Current limitations
* this API is not very extensible, but it could be easily changed to allow end users to define custom events in addition to Harmony core events
* we use a tokio broadcast channel behind the scene so only in process communication can happen, but it could be easily changed to a more flexible communication mechanism as implementation details are hidden
### `harmony_composer` VS `harmony_cli`
As Harmony Composer launches commands from Harmony (CLI), they both live in different processes. And because of this, we cannot easily make all the logging happens in one place (Harmony Composer) and get rid of Harmony CLI. At least not without introducing additional complexity such as communication through a server, unix socket, etc.
So for the time being, it was decided to preserve both `harmony_composer` and `harmony_cli` and let them independently log their stuff and handle their own responsibilities:
* `harmony_composer`: takes care only of setting up & packaging a project, delegates everything else to `harmony_cli`
* `harmony_cli`: takes care of configuring & running Harmony
### Logging & prompts
* [indicatif](https://github.com/console-rs/indicatif) is used to create progress bars and track progress within Harmony, Harmony CLI, and Harmony Composer
* [inquire](https://github.com/mikaelmello/inquire) is preserved, but was removed from `harmony` (core) as UI concerns shouldn't go that deep
* note: for now the only prompt we had was simply deleted, we'll have to find a better way to prompt stuff in the future
## Todos
* [ ] Update/Create ADRs
* [ ] Continue instrumentation for missing branches
* [ ] Allow instrumentation to emit and subscribe to custom events
Co-authored-by: Ian Letourneau <letourneau.ian@gmail.com>
Reviewed-on: https://git.nationtech.io/NationTech/harmony/pulls/91
Reviewed-by: johnride <jg@nationtech.io>
With this architecture, we have an extensible application module for which we can easily define new features and add them to application scores.
All this is driven by the ApplicationInterpret, who understands features and make sure they are "installed".
The drawback of this design is that we now have three different places to launch scores within Harmony : Maestro, Topology and Interpret. This is an architectural smell and I am not sure how to deal with it at the moment.
However, all these places where execution is performed make sense semantically : an ApplicationInterpret must understand ApplicationFeatures and can very well be responsible of them. Same goes for a Topology which provides features itself by composition (ex. K8sAnywhereTopology implements TenantManager) so it is natural for this very imp
lementation to know how to install itself.
Co-authored-by: Ian Letourneau <ian@noma.to>
Reviewed-on: https://git.nationtech.io/NationTech/harmony/pulls/70
Co-authored-by: Jean-Gabriel Gill-Couture <jg@nationtech.io>
Co-committed-by: Jean-Gabriel Gill-Couture <jg@nationtech.io>
- Adds functionality to download, install, and manage k3d clusters.
- Includes methods for downloading the latest release, creating clusters, and verifying cluster existence.
- Implements `ensure_k3d_installed`, `get_latest_release_tag`, `download_latest_release`, `is_k3d_installed`, `verify_cluster_exists`, `create_cluster` and `create_kubernetes_client`.
- Provides a `get_client` method to access the Kubernetes client.
- Includes unit tests for download and installation.
- Adds handling for different operating systems.
- Improves error handling and logging.
- Introduces a `K3d` struct to encapsulate k3d cluster management logic.
- Adds the ability to specify the cluster name during K3d initialization.
Co-authored-by: tahahawa <tahahawa@gmail.com>
Reviewed-on: https://git.nationtech.io/NationTech/harmony/pulls/9
Reviewed-by: johnride <jg@nationtech.io>
Co-authored-by: Taha Hawa <taha@taha.dev>
Co-committed-by: Taha Hawa <taha@taha.dev>
