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

Faas: Add an option to use the X-Ray env var as main parent. #263

Conversation

carlosalberto
Copy link
Contributor

This is done to offer compatibility to users with the previous behavior (which used the context in this X-Ray environment variable as the main parent).

Please review @open-telemetry/lambda-extension-approvers @dyladan @open-telemetry/python-maintainers

Note: I didn't include an actual name for this option, but I'm open to suggestions.

Closely related to open-telemetry/opentelemetry-js-contrib#1411

This is done in order to help users migrating from
the previous behavior (which is offered via this new
flag) and the new approach (which uses links).
@@ -66,6 +66,10 @@ contain an incomplete trace context which indicates X-Ray isn’t enabled. The e
`Context` will be valid and sampled only if AWS X-Ray has been enabled for the Lambda function. A user can
disable AWS X-Ray for the function if the X-Ray Span Link is not desired.

Instrumentation SHOULD include an explicit option to use `Context` in `_X_AMZN_TRACE_ID` as parent insted of `Link`,

Choose a reason for hiding this comment

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

Suggested change
Instrumentation SHOULD include an explicit option to use `Context` in `_X_AMZN_TRACE_ID` as parent insted of `Link`,
Instrumentation SHOULD include an explicit option to use `Context` in `_X_AMZN_TRACE_ID` as parent instead of `Link`,

Copy link
Contributor

Choose a reason for hiding this comment

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

It's my understanding this is a temporary stop-gap to help existing user migrate.

We should use similar language/terminology as was done in http semconv about migration period here. See: https://github.com/open-telemetry/semantic-conventions/tree/main/docs/http

Copy link
Member

Choose a reason for hiding this comment

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

Really? I think having an option to use the X-Ray context as parent seems OK also long-term. Though I would change the behavior slightly to always include the link in addition.

Copy link
Member

Choose a reason for hiding this comment

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

More than just "OK", it is imperative that X-Ray users have the ability to utilize the X-Ray span context that is provided to them by the Lambda service. We cannot have a system that makes that impossible and should not have one that leaves it to the instrumentation author to decide whether that category of users should be supported.

What benefit would be gained from having the same span context both as parent and as a link?

Copy link
Member

Choose a reason for hiding this comment

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

What benefit would be gained from having the same span context both as parent and as a link?

The benefit would be that the link would always be there independent of any instrumentation configuration.

(It could also be compared against the parent to determine if the source was X-Ray, but if that is a common use case we would be better of with a "parent_source" span attribute mirroring the link "source" attribute)

Copy link
Member

Choose a reason for hiding this comment

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

I think some of the flexibility across the semantic conventions is intentional in order to cover a broader range of (future) third-party instrumentations. I personally expect all instrumentations published under the OpenTelemetry org to follow all "Recommendeds" and SHOULDs, though I don't think we've had this discussion as a community (I've added this question to next week's semconv meeting).

I generally agree that SHOULD means "there may exist valid reasons in particular circumstances to ignore a particular item, but the full implications must be understood and carefully weighed before choosing a different course." I do not agree that there can be no other course taken as SHOULD is not MUST. I expect that maintaining compatibility while experimental specification language is being worked on to be a good reason for "choosing a different course".

Copy link
Member

Choose a reason for hiding this comment

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

I expect that maintaining compatibility while experimental specification language is being worked on to be a good reason for "choosing a different course".

👍

Copy link
Member

Choose a reason for hiding this comment

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

(at the same time, adopting the latest semconv in instrumentations is the best way to get real user feedback which can be cycled back into semconv before it gets marked as stable)

Copy link
Member

Choose a reason for hiding this comment

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

@jsuereth please see #277 for an attempt at language that would govern how instrumentation constructs a Carrier to be used with a user-specified Propagator. I believe this should address all of your concerns expressed here.

Copy link
Member

Choose a reason for hiding this comment

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

(at the same time, adopting the latest semconv in instrumentations is the best way to get real user feedback which can be cycled back into semconv before it gets marked as stable)

I think that's what happened here. The real user feedback was "this broke every existing usage of this instrumentation with no way to make it work again".

docs/faas/aws-lambda.md Outdated Show resolved Hide resolved
Copy link
Member

@dyladan dyladan left a comment

Choose a reason for hiding this comment

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

Looks good. I thought we had agreed on some sort of deadline to come up with a longer term fix. Is that not part of this PR?

CHANGELOG.md Outdated Show resolved Hide resolved
@@ -66,6 +66,10 @@ contain an incomplete trace context which indicates X-Ray isn’t enabled. The e
`Context` will be valid and sampled only if AWS X-Ray has been enabled for the Lambda function. A user can
disable AWS X-Ray for the function if the X-Ray Span Link is not desired.

Instrumentation SHOULD include an explicit option to use `Context` in `_X_AMZN_TRACE_ID` as parent insted of `Link`,
Copy link
Member

Choose a reason for hiding this comment

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

Does the above paragraph also need to be modified to be "aware" of this configuration?

@carlosalberto
Copy link
Contributor Author

@dyladan I don't think so, no. The longer term may take longer, and we don't want to block progress for now (hopefully we also have a smooth transition). Correct me if I'm wrong @open-telemetry/lambda-extension-approvers

cc @rapphil as you have a PR to address this for Python

carlosalberto and others added 2 commits August 17, 2023 12:36
Co-authored-by: Daniel Dyla <dyladan@users.noreply.github.com>
Co-authored-by: Alex Boten <alex@boten.ca>
@Aneurysm9
Copy link
Member

This change does not work for me. I have created #272 as an alternate proposal.

@@ -66,6 +66,10 @@ contain an incomplete trace context which indicates X-Ray isn’t enabled. The e
`Context` will be valid and sampled only if AWS X-Ray has been enabled for the Lambda function. A user can
disable AWS X-Ray for the function if the X-Ray Span Link is not desired.

Instrumentation SHOULD include an explicit option to use `Context` in `_X_AMZN_TRACE_ID` as parent insted of `Link`,
Copy link
Member

@Oberon00 Oberon00 Aug 21, 2023

Choose a reason for hiding this comment

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

Is it expected that instrumentations do anything special about X-Ray parents with unset sampled flag? If not, maybe a note could still be added that this may lead to all spans being unsampled unless X-Ray is enabled.

Copy link
Member

Choose a reason for hiding this comment

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

Isn't that the case with any parent span context with an unset sampled flag? If so, that's something that should probably be addressed more broadly by the spec.

Copy link
Member

@Oberon00 Oberon00 Aug 24, 2023

Choose a reason for hiding this comment

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

Yes it is, but X-Ray is unusual in that you will get a valid unsampled span context if X-Ray is disabled.

@jsuereth
Copy link
Contributor

FYI - TC discussion and opinion on this PR: #277 (comment)

@jsuereth
Copy link
Contributor

Closing this in favor of ongoing discussion in #354

@jsuereth jsuereth closed this Oct 23, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
Archived in project
Development

Successfully merging this pull request may close these issues.

7 participants