Otel collector receiver (Component?) guide #3986
Replies: 1 comment 4 replies
-
@hughesjj I agree with you that the original page needs a major rework since a lot has changed over time. What we need to discuss how much developer documentation should live in the docs and how much should be left to the repositories. In this particular case I personally lean towards having this on the website, since "building your own collector component" is something more and more people who are technically end-users will look for. They might end up in the contrib repository, but I argue that there are many use cases were folks want to have their "own" thing that might not even be OSS. A thing I'd like to call out is that we need to be clear on how many pages we want to distribute that content and how we are going to structure them. cc @open-telemetry/collector-approvers |
Beta Was this translation helpful? Give feedback.
-
In opentelemetry.io, we have an existing guide for creating a span receiver. The guide is great – to my understanding, it’s a syndicated version of a blog post made a few years ago, and we’ve done some to keep it up to date. While still useful for current and aspiring collector contributors, we haven’t given it much tender loving care over the years, and the tribal knowledge between the collector sig and what's in the document has drifted a bit.
Motivation
Anecdotally and personally, I have a few ideas for improvement that I’ve workedshopped in the community spec before along with some of my cohorts which I’d like to see us implement in the official public docs.
Most importantly, I wish for the guide to be easier to maintain for the collector-contrib SIG members and similar interested parties.
Proposal
We should break up the “creating an X” guides into a more general flow that partitions individual pages of content out into different parts of the execution cycle (requirements gathering, design, implementation, testing, acceptance, maintenance, etc), with “chapters” or subfolders that can dive into how to do so for any particular component.
The “highest level” docs could be seeding context for the various parts of the execution, leading the reader through the process for what concerns they consider. Once they have this in scope, they can dive into a detailed, guided, comprehensive tutorial for how to execute the particular task, given some arbitrary (illustrative?) choices for the aforementioned “problem context”.
Plan of attack
I’d like to do the high level first before diving into implementation, but we can do this in parallel/pipelined.
working proposal (would make these github issues etc)
Phase 1 would be additive only, in the existing
collector/building
folderPhase 1: new content
Phase 2: new format
collector/building
into a hand-held guide with flow and decision points instead of a "flat" hierarchy. Needs consensus of content first.Beta Was this translation helpful? Give feedback.
All reactions