Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

feat: Generate more telemetry via TelemetryAPI receiver #1230

Open
wants to merge 43 commits into
base: main
Choose a base branch
from

Conversation

borchero
Copy link

@borchero borchero commented Mar 31, 2024

Motivation

The current implementation of the Telemetry API receiver only generates init spans. However, it can easily be extended to create

  • spans for the entire function invocation
  • metrics for function invocations

This PR leverages the events provided by the Telemetry API to implement the latest semantic conventions for FaaS spans and metrics.

It also updates the coldstartprocessor to become a more general faasprocessor.

Please read through the READMEs of the telemetryapnireceiver and the faasprocessor for an overview of the changes.


I did not yet write unit tests (and only checked functionality manually) but wanted to check with the maintainers first whether this change would be desirable to include into this project.


This PR supersedes #1216.

@borchero borchero requested a review from a team as a code owner March 31, 2024 22:11
@tylerbenson
Copy link
Member

@rapphil I'm curious what you think about this.

"os"
"os/signal"
"path/filepath"
"sync"
"syscall"

"github.com/open-telemetry/opentelemetry-lambda/collector/lambdalifecycle"
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

How does this new component propagate context from the collector to the application?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm not sure whether this answers your question but the collector doesn't really pass any context to the application. When the application generates a span A with a parent B that was propagated from an external system, the faas processor identifies span A (by checking for spans with a faas.invocation_id) and inserts the Lambda runtime span (let's call it C) between A and B. That means, C becomes parent of A and B becomes parent of C.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks, that helps.

Copy link
Contributor

@nslaughter nslaughter left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This looks promising.

Aside from the comment I left on the ConsumeTraces method, I would be able to review this better in at least a couple of PRs. Supporting changes in the telemetryapireceiver would be the most beneficial.

Thanks for making this contribution. Looking forward to updates.

return consumer.Capabilities{MutatesData: true}
}

func (p *faasProcessor) ConsumeTraces(ctx context.Context, td ptrace.Traces) error {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Simplifying this method will help with reasoning about this code and testing paths. For example, these high-level operations make sense to me:

  • categorizeSpan (caches with runtimeSpans, initSpans, or invocationSpans)
  • associateSpans (adds traceid and parentid)
  • forwardRemainingSpans

Copy link
Contributor

@nslaughter nslaughter left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This looks promising.

Aside from the comment I left on the ConsumeTraces method, I would be able to review this better in at least a couple of PRs. Supporting changes in the telemetryapireceiver would be the most beneficial.

Thanks for making this contribution. Looking forward to updates.

Adding notes below: 2024-05-23

After some discussion, I think this would be ideal to merge as something entirely elective. I haven't worked out how to refactor this to a processor, but that would be the best path to make this available. A fallback would be to keep it a receiver as an alternative to coldstart rather than replacement.

Also note that the spirit of this review was a high-level opinion on whether the code belonged in the project and a couple of changes to facilitate a more in depth review.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

3 participants