Generate process metrics with semconv yaml

[braydonk]

Supplants #160

Changes

This PR adds a yaml version of the process metrics and changes the process metrics documentation to be generated from the yaml instead of hand-written.

Merge requirement checklist

• CONTRIBUTING.md guidelines followed.
• CHANGELOG.md updated for non-trivial changes.
• schema-next.yaml updated with changes to existing conventions.

[braydonk]

I am unable to assign reviewers, these are the folks who commented on the previous PR. @joaopgrassi @mx-psi @jsuereth

[mx-psi]

Needs minor edits for CI to pass but content LGTM

[braydonk]

Needs minor edits for CI to pass

Thanks @mx-psi those are fixed now, also added something to .yamllint to not scan node_modulues when running yamllint locally (didn't think it was worth a whole separate PR for something that small).

[braydonk]

Looks like these workflows require approval now if someone can press the button for me. 😄

[ChrsMark]

Content-wise lgtm!

@braydonk please file an issue for #330 (comment) so as to examine #260 as well :).

[braydonk]

Thanks, I'll file an issue today and I'll put my thoughts on #260 in there!

[joaopgrassi]

I will try to take a look at it today or tomorrow!

[joaopgrassi]

Here's a first pass

[braydonk]

Sorry for the force-push, I should have merged instead of rebasing to update the branch. Force of habit

[braydonk]

It's possible I'm missing some kind of context from discussions around #51, but it seems like what I have here should be fine. With the paging fault type attribute for example: it is in the process.paging.faults namespace. I added that namespace, as the attribute was previously type (just type). process.paging is not any more of a proper namespace than process.paging.faults as far as I can tell.

If I'm indeed missing some extra guidance that's not present in #51, and this attribute should go to process.paging, then I can also rename it to fault_type. But I would like to understand why this and the other pointed out attributes are not following the decision from #51 already.

[joaopgrassi]

With the paging fault type attribute for example: it is in the process.paging.faults namespace

The thing I'm not sure is that process.paging.faults is the metric, not a namespace. From reading #51 and the linked issues, my interpretation is that attributes should go on the namespace , meaning process.paging.* in this case. Maybe we need clarification if attributes can also be nested under the "implicit metric namespace".

CC @open-telemetry/specs-semconv-approvers @open-telemetry/specs-semconv-maintainers

Edit: There's also this old issue which I think is related and was never closed with a decision #50

[braydonk]

I have updated the PR to match the semantic convention guidance as well as I'm able.

I renamed the metrics that are updowncounters to use the .count pluralization method. I also changed the metric attributes such that they do not use the metric names as namespaces.

[braydonk]

I added proposed_changelog.md and proposed_schema_next.yaml, which upon PR approval I will add properly to the CHANGELOG.md and schema-next.yaml files. This is to avoid constantly needing to rebase these giant files if the PR lasts over a long period of time.

Do not merge until I've added these back to the main files.

[joaopgrassi]

LGTM after fixing the CI issues

[ChrsMark]

LGTM. Let's make sure that we file an issue for moving the attributes to the registry: #330 (comment)

[braydonk]

LGTM. Let's make sure that we file an issue for moving the attributes to the registry: #330 (comment)

Opened #650 to track this.

[braydonk]

This should be ready for merge.

[joaopgrassi]

I will bring this up in the semconv SIG meeting today, to raise awareness. I guess we can merge it then later today/tomorrow if no objections.