"example" project is not helpful for beginners about grafana-dashboard HOT 1 CLOSED

Thank you for your comments!

The example cluster was meant to be run with docker. There is a README section with short description https://github.com/tarantool/grafana-dashboard#experimental-cluster . If you use docker-compose up, it will start up containers and a network which will make up configured and fully functional monitoring stack (even two of them) connected to a Tarantool example app. Tarantool container is a container which runs luatest that starts up a cluster, configures it and send some requests to make up some load for graphs (https://github.com/tarantool/grafana-dashboard/tree/master/example/project/cluster). Metrics output is configured with clusterwide configuration with luatest, and that is the reason why we don't have metrics.set_export code for now. The other reason is that this cluster was introduced before set_export and there wasn't any obvious reasons why we should upgrade. The cluster runs with luatest instead of cartridge-cli because it was built before introduction of replicasets.yml and it was not convenient to works with cluster as I wanted to (for example, set up roles and replicasets automatically).

So if you want to use this example cluster to test only Grafana-related functionality, you can simply run docker-compose up and go to localhost:3000 to set up some boards. You do not need to set up Prometheus or Tarantool metrics output. If you want to change example app code, you can change example app code in folders and, if needed, changed enabled roles or cluster topology here if you're familiar with luatest https://github.com/tarantool/grafana-dashboard/tree/master/example/project/cluster (you may need to run docker-compose build --no-cache to reload code if you're have started a cluster before). If you want to monitor some app that already runs and have exposed metrics on your localhost, you may also use docker containers (Prometheus + Grafana) but configure your instances host:port in Prometheus config.

The reason that configuration uses example_project: is also so we could run it with docker-compose (which build a network for containers) and it will set up without any manual configuration.

It was planned to migrate to using cartridge-cli instead of luatest (#34) someday to set up example cluster. Maybe it is the time. But I still plan to run in inside docker container for now. The main reason for example cluster is to create an environment in which it is convenient to develop a dashboard. You can call a single docker-compose up and then go to localhost:3000 to check up your new panel. So we won't change target urls until something changes.

Finally, we have a documentation page on how to import a dashboard and set up parameters https://www.tarantool.io/en/doc/latest/book/monitoring/grafana_dashboard/ . I have a link in README, but it seems like it's easy to miss. And looks like it is still not obvious how to find a job from here.

I hope that I can use this issue to build up such a README so its crucial points will be hard to miss.

from grafana-dashboard.