Telegraf config


#1

Hi guys -

I’m trying to configure Telegraf and Grafana for Linekrd. Using the exec input plugin as mentioned in Linkerd-examples, it seems to work and indeed query the proper endpoint.

From the other hand, i’ve imported the Grafana dashboard from that repository as well, and I think I have a lack of understanding in the schema design of the metrics’ outputting for Influx, since the dashboards expects certain tags I just don’t have. In the examples, the tags used are l5d_tag1 = "foo" - what’s that supposed to mean for me? Should I replace with… host name (supposed to be injected with the curl query though)? Or something else? Now I have only general tags that I’ve configured for many other plugins (env, host, dc, etc.) but no tags related to Linkerds client and routers and so on, as I’ve seen in the Grafana dashboard templates. How can I achive a per-service and per-router tagging?

edit
Now I understand that actually I don’t have enough measurements. Somehow, with my config, telegraf outputting all the linkerd-related metrics to a single measurement we called “linkerd”, in our single database for telegraf collection. So if I run show measurements inside the metrics database (after use <database>, I mean), I have only one measurement for linkerd. Is this expected?


#2

Hi @jacobgo. One measurement is not expected. Were you able to get the example working as described in https://github.com/linkerd/linkerd-examples/tree/master/influxdb ?

With that example, you should see all stats prefixed with l5d:. The l5d_tag* are just examples, not necessary unless you need an additional dimension to query your data.

Example list of measurements:

Example list of tags: