Describe the bug
Some tracing vendors use vendor-specific trace headers which take precedence over the W3C trace context headers. Those are propagated as they are and cause trace propagation issues at the upstream service because that one does not pick up updated trace headers.
To reproduce
Steps to reproduce the behavior:
- Set up and run a demo application consisting of (client app, nginx, and server-app) that is using vendor-specific trace headers and nginx-otel in the middle in the observability vendor environment.
- Check trace correlation on the vendor UI
- tcpdump the demo application
Expected behavior
There is a configuration option for nginx-otel to drop unwanted headers from the outbound HTTP request.
Your environment
Additional context
With OpenTracing this was way easier but the data structures inside of nginx-otel are too limited for this.
Setting those headers to be dropped to empty string is not enough. They need to be dropped completely.
Describe the bug
Some tracing vendors use vendor-specific trace headers which take precedence over the W3C trace context headers. Those are propagated as they are and cause trace propagation issues at the upstream service because that one does not pick up updated trace headers.
To reproduce
Steps to reproduce the behavior:
Expected behavior
There is a configuration option for nginx-otel to drop unwanted headers from the outbound HTTP request.
Your environment
Additional context
With OpenTracing this was way easier but the data structures inside of nginx-otel are too limited for this.
Setting those headers to be dropped to empty string is not enough. They need to be dropped completely.