Tracing without performance

[ribice]

Continuation from #752

[ribice]

@cleptric There was a comment on old type to use this for transactions as well?

[cleptric]

We have to throw this on all events, so check-ins, errors, transactios, etc.

[codecov]

Codecov Report

Attention: Patch coverage is 89.43089% with 13 lines in your changes are missing coverage. Please review.

Project coverage is 81.92%. Comparing base (23c5137) to head (9f80641).

Files	Patch %	Lines
propagation_context.go	86.66%	3 Missing and 3 partials ⚠️
scope.go	84.84%	4 Missing and 1 partial ⚠️
hub.go	84.61%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #833      +/-   ##
==========================================
+ Coverage   81.80%   81.92%   +0.12%     
==========================================
  Files          50       51       +1     
  Lines        4436     4543     +107     
==========================================
+ Hits         3629     3722      +93     
- Misses        660      669       +9     
- Partials      147      152       +5     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ribice]

The thing I feel wrong about is that I don't get the traces nested. Am I doing something wrong here?

The parent span ID is set properly.

[aldy505]

The thing I feel wrong about is that I don't get the traces nested. Am I doing something wrong here?

The parent span ID is set properly.

@ribice can I see the code to repro this one?

[ribice]

The thing I feel wrong about is that I don't get the traces nested. Am I doing something wrong here?
The parent span ID is set properly.

@ribice can I see the code to repro this one?

https://gist.github.com/ribice/2c583f9c1edc3084d3572e2ea5065419

[aldy505]

The thing I feel wrong about is that I don't get the traces nested. Am I doing something wrong here?
The parent span ID is set properly.

@ribice can I see the code to repro this one?

https://gist.github.com/ribice/2c583f9c1edc3084d3572e2ea5065419

I think your sentry.ContinueTrace is conflicting with the ones from here on the sentryhttp middleware

sentry-go/http/sentryhttp.go

Lines 113 to 122 in 9f80641

	options := []sentry.SpanOption{
	sentry.WithOpName("http.server"),
	sentry.ContinueFromRequest(r),
	sentry.WithTransactionSource(sentry.SourceURL),
	}
	transaction := sentry.StartTransaction(ctx,
	fmt.Sprintf("%s %s", r.Method, r.URL.Path),
	options...,
	)

The sentry.ContinueFromRequest should acquire the Sentry-related headers and append it to the span options. Try to remove that part from the middleware and see if everything works, maybe?

[aldy505]

Got time to finally come back to this, made a repro on Nodejs, nothing wrong about it not being nested. But I wonder why it got the same parent span id though. It should only got 2-3 captured events

Repro https://gist.github.com/aldy505/90354abc6877071dec601c8ddabca092

[ribice]

Hmm, seems you're right. Capturing it first time works properly (same as in node snippet above), but every subsequent request appends to that trace which is weird. I'll investigate now.

First:

Second:

[ribice]

I was able to resolve it by creating a new Scope/Hub in every request. This way it generates new traceID/spanID.

https://gist.github.com/ribice/2c583f9c1edc3084d3572e2ea5065419 (service1-updated.go).

Not sure if this is the intended way.

[aldy505]

On the JS SDK, the ContinueTrace is a function that accepts a callback. I think on the callback, it creates a new scope internally, Not sure if this should also be done on Go though.

Wanted to comment