Fluentd persistence of logs in Elasticsearch and of config of Kibana

[stygmate]

Summary

adding parameters to enable script to define persistence parameters for ES and Kibana data.

Why is this important?

loosing all logs in case of rescheduling is not ok in production.

Are you interested in contributing to this feature?

maybe, if i have time.

Question:

What is the best way to configure persistent storage for now ?

[madalinignisca]

ELK will use default storage class. This is not the responsibility of the ELK addon. Any Kubernetes administrator would setup storage and define a default class before anything else; I assume that who originally created the addon made this assumption.

[stygmate]

ELK will use default storage class. This is not the responsibility of the ELK addon. Any Kubernetes administrator would setup storage and define a default class before anything else; I assume that who originally created the addon made this assumption.

the creator of the addon use emptydir instead of pvc so default storage driver don't have anything to do with this. https://github.com/canonical/microk8s-community-addons/blob/main/addons/fluentd/fluentd/es-statefulset.yaml#L115

[madalinignisca]

@stygmate, auch, my bad, did not even looked over it, as I have my own setup for ELK. You are right. My first question that hit me was "WHYYY?".

I now think that addons must follow some conventions. Example, addons should support out of the box default storage class, or allow by some params to use a user given class.