You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
One of these days we should take a swing at adding otel metrics to otel-cli. I don't know if it'll work out but it feels worth exploring.
e.g. (having not dug into metrics at all)
otel-cli counter foo.bar 1
otel-cli histo ???????
Should there be another layer of subcommands? e.g. otel-cli trace span -n "my span"? It feels heavy to me and I'm curious what others think.
Reacted by Derek Murawsky, Ryan Dens, Ishwor Gurung, Uri Savelchev, robertaves, Steven Swartz, Zac Pullar-Strecker, Ivan Kurchenko, Jannis, Joe Reuter and 2 more
Welllll we should probably wait until support for metrics gets a little better - I think the spec is just now about to be finalized (and support is pretty non-existent across the ecosystem).
But - I think yes that would be intriguing to explore, and could interesting!
Should there be another layer of subcommands? e.g. otel-cli trace span -n "my span"? It feels heavy to me and I'm curious what others think.
I agree, that feels heavy. That's not to say that we can't explore different schemes for the commands, but I do tend to dislike structures that are more than one level deep.
Has there been any update on the metrics spec front? This would be a great feature to have.
Currently, we use curl to send metrics out of our build pipeline to InfluxDB via their simple line protocol. We're looking at adding OpenTelemetry to the CI/CD system as a whole, and having the ability to push metrics out via a CLI would greatly simplify migration. Bonus points for tagging the metrics to the related span that they occurred in.
Reacted by Ishwor Gurung, ET, Zac Pullar-Strecker, Ivan Kurchenko and Kevin Pullin
Activity
ahayworth commentedon Aug 19, 2021
Welllll we should probably wait until support for metrics gets a little better - I think the spec is just now about to be finalized (and support is pretty non-existent across the ecosystem).
But - I think yes that would be intriguing to explore, and could interesting!
I agree, that feels heavy. That's not to say that we can't explore different schemes for the commands, but I do tend to dislike structures that are more than one level deep.
derekmurawsky commentedon Apr 19, 2022
Has there been any update on the metrics spec front? This would be a great feature to have.
Currently, we use curl to send metrics out of our build pipeline to InfluxDB via their simple line protocol. We're looking at adding OpenTelemetry to the CI/CD system as a whole, and having the ability to push metrics out via a CLI would greatly simplify migration. Bonus points for tagging the metrics to the related span that they occurred in.