Plugins conf in general, grafana in particular

I want to give a try to grafana/prometheus. Installing them on microk8s is really easy but changing the configuration seems much less easy. I noticed other posts that when unanswered, and this baffles and scares me.

When installing Traefik I stumbled on the same problem and at last, I decided to install via helm, pretty easy and is very easy to further customize.

Same was for fluentd whose default configuration doesn’t persist

Which is the suggested way to cope with the configuration of these plugins? are there people that use the prepacked version? should I go with grafana-operator via helm? (or should I use grafana via helm…)

The single first easy thing for me is to simply configure the smtp, it should be a simple env variable… but how to add an env variable is the installation is just 'microk8s enable Prometheus? Should I patch the deployment? Then I should restart the pod… how to set persistence?

Any hints is appreciated



IMHO, as it currently stands the addons do not have plenty of knobs to easily customize them.

Hence end up just getting to the upstream, customize it, then apply to the cluster.

The prometheus addon is using the kube-prometheus.


Thanks @balchua1,
names of the services are a little bit different from what you get from helm installation. Configuration as well.
it seems to me -but I’m definitely not an expert- that would be more useful to be able to configure services that one installs via plugin. Wouldn’t it be more useful if behind the scenes "microk8s enable " was really a “helm install -f values.yaml” with known values so that if would be very easy to fin tune it according to our needs?

With regards to the prometheus addon, it is not using helm. Its plain manifest taken from the lube-prometheus default generated set of manifests. The kube-prometheus actually uses jsonet to generate the manifests.