Agent 7.71.0 breaks OTLP metrics ingestion for .NET System.Runtime metrics

[lance-cognitiv]

After upgrading from Datadog Agent 7.70 to 7.71.0, OTLP metrics from .NET applications stopped appearing in Datadog, despite being successfully sent and received by the agent.
Environment:
Datadog Agent: 7.71.0
Platform: Ubuntu on AWS EC2
.NET Version: 9.0
OpenTelemetry: 1.11.2
OTLP Protocol: HTTP/protobuf on port 4318
Symptoms:
.NET System.Runtime metrics (dotnet.gc.collections, dotnet.monitor.lock_contentions, etc.) are not appearing in Datadog
tcpdump confirms metrics ARE being sent to port 4318 and received by the agent
OTLP connection shows as ESTABLISHED
Agent status shows "OTLP Collector status: Running"
No errors in agent logs
Reverting to Agent 7.70.0 immediately resolves the issue

[liustanley]

Hi @lance-cognitiv, thanks for reporting this issue. I was able to confirm this same behavior and the root cause is that Agent v7.71.0 is using collector v0.133.0 which includes this bug: open-telemetry/opentelemetry-collector#13727. The next Agent version will include a collector version that fixes this bug.

If you need a workaround in the meantime, you can either use gRPC to receive OTLP data or downgrade your Agent version to v7.70.0.

[lance-cognitiv]

Thanks for the update!