loki-operator

README

Loki operator

This operator is expected to gather partial promtail configurations relevant to specific applications and combine them into the single promtail config file in its ConfigMap.

Warning: This is proof-of-concept quality software! Don't run in production until you really YOLO!

How it works

Each application can deliver its own promtail log parsing config by creating a ConfigMap in the same namespace. The name of the ConfigMap is arbitrary. The ConfigMap needs to have just 1 key promtail.yaml, which includes promtail config to include in the actual promtail ConfigMap on behalf of this application.

The application informs loki-operator to register its config by including the Label in Pod's template yaml (can be single Pod or any Pod created by Deployment or any other controller). The Label looks like this

giantswarm.io/loki-promtail-config: apiserver-promtail-config

If there are more than 1 container in the Pod (like because you're running some kind of service mesh), you have to explicitly point to the container you want to get logs from:

giantswarm.io/loki-promtail-container: apiserver

What's missing

• any tests
• almost all validation of input files
• currently the application needs to provide both the actual configuration pipeline but also filters config to apply the pipeline only to valid pods; this second part (where to apply the config - the filter) should be auto-generated
• restarts of promtail pods
• ability to get logs from many containers of the same Pod applying different configs to them.