Details
-
New Feature
-
Resolution: Done
-
Major
-
None
-
None
Description
For those running with containers already, a distributed rather than single container option would be a better choice. We can iterate the components independently then quite easily.
This could be provided as a Helm chart or similar so we just reuse the PLG helm chart and add our customisation plus additional components.
This can still be deployed on-premise as well, just as a set of containers (e.g. docker compose stack) rather than a single one. It would open up options for scaling and persisting independent components, plus is the recommendation for deploying the PLG stack in production: https://grafana.com/docs/loki/latest/overview/#modes-of-operation
It would be better as well to allow us to pick up the official images directly, particularly for Fluent Bit which does not support Alpine by default.
Per Patrick Stephens: I think we should also support a more granular deployment anyway for both on-premise and kubernetes deployments: docker compose stack and helm charts can target a local deployment. This would allow better control of the individual components and scaling, plus upgrading, etc. is easier.
Still provide a microlith but also a local multi-container stack using the official documented ways to do that for the Grafana products and our own healthcheck.
Attachments
Issue Links
- duplicates
-
CMOS-305 Distributed deployment, scaling, and high availability
- To Do