simonschneider / monarch Goto Github PK

in the list of existing resources color code them depending on if they are used or not (maybe by whom etc) to help configuration. Do this in real time as the form is being filled out.

With github actions
on every commit
Linting, tests, docker image build, helm chart testing.

on releases
push docker image to docker hub and release helm chart

A UI where to configure monarch

both backend and frontend.
Jest, supertest, fetch-mock etc.

The lib would have to be updated to support grouping/clustering via the dot graph clusters:

https://graphviz.org/Gallery/directed/cluster.html

this could be used to group multiple services into clusters that could help visualize domain boundaries etc..

Eslint for both backend and frontend.

Allow templates in queries that are provided in the frontend

A timeline with a window selector to drag to select an aggregation window which will be applied to the graph.

It would be possible to move large parts of the service clientside and only have the express server as a configuration store and prometheus proxy.

This would allow the possibility to run a demo without installing anything.

the UI would have some check if it can reach a server where it is hosted, if it can't it will gracefully degrade into a serverless version.
the Serverless version will save the configuration file in redux only and not sync it to the backend.
The client would also, regardless of servless or not, have a configuration option to either request prometheus metrics from it's own proxy or from whatever user specified prometheus url.

In the server version the only thing different would be that the client would retrieve and persist configuration to the backend.

This requires all the business logic to move to the frontend but it makes the server extremely simple and slim. It would be used to serve the frontend, proxy prometheus requests and act as a configuration store.

This could be visualized in some good way (and also be fed as a critical/warning symbol)
perhaps visualize with stacking of cards etc...

Add RED request metrics

Create a form that helps produce the JSON needed to configure the backend.

Helm chart for easier deployment

MVP for configurability.

just allow a new configuration to be posted and the current to be downloaded.

If a consumer is lagging over a threshold the CG edge from the topic to the service is marked with warning (yellow).

If a consumer is below a certain percentage of consumption rate vs production rate the CG edge from the topic to the service is marked with critical (red)

Download configuration in JSON format

This could be to distinguish what topics can be used to integrate to services

Right now MonArch crashes on:

• missing topics (services producing or consuming from non-existing topics)
• missing services (services requesting from non-existing other services)
• missing consumergroups (services consuming from non-existent CGs)

Produce lists of (via metrics)

• found services
• found topics
• found CGs

show these in a new visual configuration tab.

Support integrating to docs-prox to add a link to the services documentation

Create a sidebar with controls and a main content page with everything else.
The sidebar should be minimal to take away as little as possible from the actual visualization