The current RH documentation is missing any "good practices" for the collector and the LokiStack.

What's expected is to have some guidelines about the collector and LokiStack.

Some examples for the collector:

• How to isolate the inputs from the outputs and the impact of this (if two inputs are reading the same logs, then, it's read twice the same logs from disk)
• Set the limits and requests and how to calculate them
• How to filter for reducing the size of the logs (read/sent)
• How to set up the pipelines taking into consideration the existent filters
• etc

For LokiStack:

• How to measure to take the correct size: number of GB read
• When to scale up the pods per component or when move to the next Loki Size
• Increase the number of replicas for the Ingester for supporting that one Ingester is down
• Recommendation for how to set up Loki with ODF: RGW or noobaa?
• Run the LokiStack pods in separate nodes as the pods has not limits.cpu and limits.memory and for the impact in the resources in the nodes that it has
• How to tune the querier and the Logging UI timeouts for not being the default of 30s