Zipkin + Prometheus + Grafana = <3

1,559 views

Skip to first unread message

Zoltán Nagy

unread,

Mar 1, 2017, 2:32:50 PM3/1/17

to zipkin-user

Hey all,

Part of the production story of Zipkin is operations and monitoring of Zipkin itself. For a while now Zipkin has directly exposed a bunch of metrics in the Prometheus format (on the /prometheus endpoint, in case you were wondering). The next step is to provide an example Prometheus setup and a dashboard to consume that data.

https://github.com/openzipkin/docker-zipkin/pull/135 aims to bring in Prometheus and Grafana into the default docker-compose.yml file in docker-zipkin. https://grafana.net/dashboards/1598 (currently in draft, not listed in search results) then aims to provide a sane-defaults, well-designed dashboard to visualize all that data. I'm looking for help, on two fronts.

Is the dashboard "correct"? As in, are the labels saying what the metrics actually are? I built it to the best of my ability, but I'd love more eyes on it. Also, are the descriptions easy to understand? Do the graphs make sense? Would other groupings of data series make more sense?
(Ideally after 1. is taken care of) If you happen to already run Zipkin + Prometheus in production, I'd love to upload a screenshot of the dashboard with actual, real data - the current screenshot is just the idling example docker-compose setup.