Service call chain broken based on gRPC protocol

[yitian108]

Describe the bug

When I used OpenTelemetry go instrumentation to perform automatic instrumentation on the GRPC program to verify the client->middle->middle->server call chain, three services based on k8s were launched on December 11, 2025 at T08:51:03. From the visualization of the call chain from Tempo, it can be seen that the call logic from 08:51:03 to 08:53:33 basically conforms to the client->middle->middle->server call chain. However, from 08:53:33 onwards, each trace on the client side only has one span of its own, and there is no longer a middle or server side. Moreover, some traces from 08:51:03 to 08:53:33 have duplicate middle spans.

Environment

• OS: AlmaLinux 9.6
• Go Version: 1.24.10
• Version: otel/autoinstrumentation-go:latest (v0.23.0)

To Reproduce

Steps to reproduce the behavior:

1. kubectl apply -f grpc-auto-demo.yaml
2. select grpc-auto-client-svc/grpc-auto-middle-svc/grpc-auto-server-svc
3. query and inspect each traceID
4. See only one span for client and no other spans for middle/server after startup a few minutes

Expected behavior

should be a full call chains client->middle->middle->server all the time rather than only lasting a few minutes from starting the pods

Additional context

please see the attachments

go mods.txt
grpc-auto-demo.yaml
otel-logs.log