feat(instrumentation): Add gRPC instrumentation

[kakkoyun]

Signed-off-by: Kemal Akkoyun kemal.akkoyun@datadoghq.com

[github-actions]

📊 Semantic Convention Registry Update Check

Current Project Version: v1.37.0 (from .semconv-version)

ℹ️ This is a non-blocking informational check that shows what semantic convention updates are available. It helps you stay informed about new conventions without requiring immediate action.

🆕 Available Updates

Comparing latest (main branch) vs v1.37.0 (current)

💡 New semantic conventions are available. Consider updating if these changes are relevant to your instrumentation:

📋 View available updates

Summary of Schema Changes

Registry versions

Baseline: unversioned

Head: unversioned

Registry Attributes

Added

• app.screen.id
• app.screen.name
• gcp.apphub_destination.application.container
• gcp.apphub_destination.application.id
• gcp.apphub_destination.application.location
• gcp.apphub_destination.service.criticality_type
• gcp.apphub_destination.service.environment_type
• gcp.apphub_destination.service.id
• gcp.apphub_destination.workload.criticality_type
• gcp.apphub_destination.workload.environment_type
• gcp.apphub_destination.workload.id
• gen_ai.embeddings.dimension.count
• gen_ai.evaluation.explanation
• gen_ai.evaluation.name
• gen_ai.evaluation.score.label
• gen_ai.evaluation.score.value
• gen_ai.tool.call.arguments
• gen_ai.tool.call.result
• gen_ai.tool.definitions
• jsonrpc.protocol.version
• jsonrpc.request.id
• k8s.pod.hostname
• k8s.pod.ip
• k8s.pod.start_time
• k8s.pod.status.phase
• k8s.pod.status.reason
• nfs.operation.name
• nfs.server.repcache.status
• onc_rpc.procedure.name
• onc_rpc.procedure.number
• onc_rpc.program.name
• onc_rpc.version
• openshift.clusterquota.name
• openshift.clusterquota.uid
• otel.event.name
• pprof.location.is_folded
• pprof.mapping.has_filenames
• pprof.mapping.has_functions
• pprof.mapping.has_inline_frames
• pprof.mapping.has_line_numbers
• pprof.profile.comment
• process.context_switch.type
• process.state
• rpc.request.metadata
• rpc.response.metadata
• rpc.response.status_code
• rpc.system.name
• system.memory.linux.slab.state
• system.paging.fault.type

Renamed

• system.processes.status --> process.state (Note: Replaced by process.state.)
• rpc.connect_rpc.error_code --> rpc.response.status_code (Note: Replaced by rpc.response.status_code.)
• system.paging.type --> system.paging.fault.type (Note: Replaced by system.paging.fault.type.)
• process.paging.fault_type --> system.paging.fault.type (Note: Replaced by system.paging.fault.type.)
• rpc.jsonrpc.request_id --> jsonrpc.request.id (Note: Replaced by jsonrpc.request.id.)
• rpc.system --> rpc.system.name (Note: Replaced by rpc.system.name.)
• rpc.connect_rpc.response.metadata --> rpc.response.metadata (Note: Replaced by rpc.response.metadata.)
• linux.memory.slab.state --> system.memory.linux.slab.state (Note: Replaced by system.memory.linux.slab.state.)
• rpc.jsonrpc.version --> jsonrpc.protocol.version (Note: Replaced by jsonrpc.protocol.version.)
• system.cpu.logical_number --> cpu.logical_number (Note: Replaced by cpu.logical_number.)
• process.context_switch_type --> process.context_switch.type (Note: Replaced by process.context_switch.type.)
• rpc.connect_rpc.request.metadata --> rpc.request.metadata (Note: Replaced by rpc.request.metadata.)
• system.process.status --> process.state (Note: Replaced by process.state.)
• rpc.grpc.request.metadata --> rpc.request.metadata (Note: Replaced by rpc.request.metadata.)
• rpc.grpc.response.metadata --> rpc.response.metadata (Note: Replaced by rpc.response.metadata.)

Uncategorized

• db.cosmosdb.status_code (Note: Use db.response.status_code instead.)
• db.redis.database_index (Note: Uncategorized.)
• enduser.role (Note: Use user.roles instead.)
• event.name (Note: The value of this attribute MUST now be set as the value of the EventName field on the LogRecord to indicate that the LogRecord represents an Event.
  )
• http.flavor (Note: Split into network.protocol.name and network.protocol.version)
• messaging.kafka.destination.partition (Note: Record string representation of the partition id in messaging.destination.partition.id attribute.)
• rpc.grpc.status_code (Note: Use string representation of the gRPC status code on the rpc.response.status_code attribute.)
• rpc.jsonrpc.error_code (Note: Use string representation of the error code on the rpc.response.status_code attribute.)
• rpc.jsonrpc.error_message (Note: Use the span status description or error.message attribute on other signals.)

Metrics

Added

• container.memory.available
• container.memory.paging.faults
• container.memory.rss
• container.memory.working_set
• k8s.container.cpu.limit_utilization
• k8s.container.cpu.request_utilization
• k8s.cronjob.job.active
• k8s.daemonset.node.current_scheduled
• k8s.daemonset.node.desired_scheduled
• k8s.daemonset.node.misscheduled
• k8s.daemonset.node.ready
• k8s.deployment.pod.available
• k8s.deployment.pod.desired
• k8s.hpa.pod.current
• k8s.hpa.pod.desired
• k8s.hpa.pod.max
• k8s.hpa.pod.min
• k8s.job.pod.active
• k8s.job.pod.desired_successful
• k8s.job.pod.failed
• k8s.job.pod.max_parallel
• k8s.job.pod.successful
• k8s.node.cpu.allocatable
• k8s.node.ephemeral_storage.allocatable
• k8s.node.memory.allocatable
• k8s.node.memory.available
• k8s.node.memory.paging.faults
• k8s.node.memory.rss
• k8s.node.memory.working_set
• k8s.node.pod.allocatable
• k8s.pod.memory.available
• k8s.pod.memory.paging.faults
• k8s.pod.memory.rss
• k8s.pod.memory.working_set
• k8s.pod.status.phase
• k8s.pod.status.reason
• k8s.replicaset.pod.available
• k8s.replicaset.pod.desired
• k8s.replicationcontroller.pod.available
• k8s.replicationcontroller.pod.desired
• k8s.statefulset.pod.current
• k8s.statefulset.pod.desired
• k8s.statefulset.pod.ready
• k8s.statefulset.pod.updated
• nfs.client.net.count
• nfs.client.net.tcp.connection.accepted
• nfs.client.operation.count
• nfs.client.procedure.count
• nfs.client.rpc.authrefresh.count
• nfs.client.rpc.count
• nfs.client.rpc.retransmit.count
• nfs.server.fh.stale.count
• nfs.server.io
• nfs.server.net.count
• nfs.server.net.tcp.connection.accepted
• nfs.server.operation.count
• nfs.server.procedure.count
• nfs.server.repcache.requests
• nfs.server.rpc.count
• nfs.server.thread.count
• openshift.clusterquota.cpu.limit.hard
• openshift.clusterquota.cpu.limit.used
• openshift.clusterquota.cpu.request.hard
• openshift.clusterquota.cpu.request.used
• openshift.clusterquota.ephemeral_storage.limit.hard
• openshift.clusterquota.ephemeral_storage.limit.used
• openshift.clusterquota.ephemeral_storage.request.hard
• openshift.clusterquota.ephemeral_storage.request.used
• openshift.clusterquota.hugepage_count.request.hard
• openshift.clusterquota.hugepage_count.request.used
• openshift.clusterquota.memory.limit.hard
• openshift.clusterquota.memory.limit.used
• openshift.clusterquota.memory.request.hard
• openshift.clusterquota.memory.request.used
• openshift.clusterquota.object_count.hard
• openshift.clusterquota.object_count.used
• openshift.clusterquota.persistentvolumeclaim_count.hard
• openshift.clusterquota.persistentvolumeclaim_count.used
• openshift.clusterquota.storage.request.hard
• openshift.clusterquota.storage.request.used
• process.unix.file_descriptor.count
• process.windows.handle.count
• rpc.client.call.duration
• rpc.server.call.duration
• system.memory.linux.available
• system.memory.linux.slab.usage
• system.network.dropped
• system.network.packets
• v8js.memory.heap.space.available_size
• v8js.memory.heap.space.physical_size

Renamed

• process.open_file_descriptor.count --> process.unix.file_descriptor.count (Note: Replaced by process.unix.file_descriptor.count.)
• k8s.daemonset.desired_scheduled_nodes --> k8s.daemonset.node.desired_scheduled (Note: Replaced by k8s.daemonset.node.desired_scheduled.)
• k8s.hpa.current_pods --> k8s.hpa.pod.current (Note: Replaced by k8s.hpa.pod.current.)
• v8js.heap.space.available_size --> v8js.memory.heap.space.available_size (Note: Replaced by v8js.memory.heap.space.available_size.)
• k8s.deployment.available_pods --> k8s.deployment.pod.available (Note: Replaced by k8s.deployment.pod.available.)
• k8s.cronjob.active_jobs --> k8s.cronjob.job.active (Note: Replaced by k8s.cronjob.job.active.)
• k8s.daemonset.current_scheduled_nodes --> k8s.daemonset.node.current_scheduled (Note: Replaced by k8s.daemonset.node.current_scheduled.)
• k8s.node.allocatable.memory --> k8s.node.memory.allocatable (Note: Replaced by k8s.node.memory.allocatable.)
• k8s.job.max_parallel_pods --> k8s.job.pod.max_parallel (Note: Replaced by k8s.job.pod.max_parallel.)
• k8s.node.allocatable.ephemeral_storage --> k8s.node.ephemeral_storage.allocatable (Note: Replaced by k8s.node.ephemeral_storage.allocatable.)
• k8s.job.failed_pods --> k8s.job.pod.failed (Note: Replaced by k8s.job.pod.failed.)
• k8s.replicationcontroller.desired_pods --> k8s.replicationcontroller.pod.desired (Note: Replaced by k8s.replicationcontroller.pod.desired.)
• k8s.hpa.desired_pods --> k8s.hpa.pod.desired (Note: Replaced by k8s.hpa.pod.desired.)
• k8s.replication_controller.desired_pods --> k8s.replicationcontroller.pod.desired (Note: Replaced by k8s.replicationcontroller.pod.desired.)
• k8s.node.allocatable.pods --> k8s.node.pod.allocatable (Note: Replaced by k8s.node.pod.allocatable.)
• k8s.job.desired_successful_pods --> k8s.job.pod.desired_successful (Note: Replaced by k8s.job.pod.desired_successful.)
• k8s.daemonset.misscheduled_nodes --> k8s.daemonset.node.misscheduled (Note: Replaced by k8s.daemonset.node.misscheduled.)
• k8s.job.active_pods --> k8s.job.pod.active (Note: Replaced by k8s.job.pod.active.)
• k8s.statefulset.updated_pods --> k8s.statefulset.pod.updated (Note: Replaced by k8s.statefulset.pod.updated.)
• k8s.statefulset.desired_pods --> k8s.statefulset.pod.desired (Note: Replaced by k8s.statefulset.pod.desired.)
• k8s.statefulset.ready_pods --> k8s.statefulset.pod.ready (Note: Replaced by k8s.statefulset.pod.ready.)
• v8js.heap.space.physical_size --> v8js.memory.heap.space.physical_size (Note: Replaced by v8js.memory.heap.space.physical_size.)
• k8s.replication_controller.available_pods --> k8s.replicationcontroller.pod.available (Note: Replaced by k8s.replicationcontroller.pod.available.)
• k8s.hpa.max_pods --> k8s.hpa.pod.max (Note: Replaced by k8s.hpa.pod.max.)
• system.linux.memory.slab.usage --> system.memory.linux.slab.usage (Note: Replaced by system.memory.linux.slab.usage.)
• k8s.daemonset.ready_nodes --> k8s.daemonset.node.ready (Note: Replaced by k8s.daemonset.node.ready.)
• k8s.deployment.desired_pods --> k8s.deployment.pod.desired (Note: Replaced by k8s.deployment.pod.desired.)
• k8s.replicaset.available_pods --> k8s.replicaset.pod.available (Note: Replaced by k8s.replicaset.pod.available.)
• k8s.node.allocatable.cpu --> k8s.node.cpu.allocatable (Note: Replaced by k8s.node.cpu.allocatable.)
• k8s.hpa.min_pods --> k8s.hpa.pod.min (Note: Replaced by k8s.hpa.pod.min.)
• k8s.replicaset.desired_pods --> k8s.replicaset.pod.desired (Note: Replaced by k8s.replicaset.pod.desired.)
• k8s.statefulset.current_pods --> k8s.statefulset.pod.current (Note: Replaced by k8s.statefulset.pod.current.)
• system.linux.memory.available --> system.memory.linux.available (Note: Replaced by system.memory.linux.available.)
• k8s.job.successful_pods --> k8s.job.pod.successful (Note: Replaced by k8s.job.pod.successful.)
• k8s.replicationcontroller.available_pods --> k8s.replicationcontroller.pod.available (Note: Replaced by k8s.replicationcontroller.pod.available.)

Obsoleted

• rpc.client.requests_per_rpc (Note: Removed, no replacement at this time.)
• rpc.client.responses_per_rpc (Note: Removed, no replacement at this time.)
• rpc.server.requests_per_rpc (Note: Removed, no replacement at this time.)
• rpc.server.responses_per_rpc (Note: Removed, no replacement at this time.)

Uncategorized

• rpc.client.duration (Note: Replaced by rpc.client.call.duration with unit s.)
• rpc.server.duration (Note: Replaced by rpc.server.call.duration with unit s.)

Events

Added

• gen_ai.evaluation.result

---

📝 What This Means

• Non-blocking: This check will never fail your PR
• Informational: Shows what's new in the semantic conventions
• Optional: You can choose when to adopt new conventions

How to Update (if desired):

1. Review the changes above to see if they're relevant to your use case
2. Update Go imports in pkg/instrumentation/**/semconv/ to the new semconv version
3. Update the version in .semconv-version file
4. Run make registry-check locally to validate

Generated by OTel Weaver • Run make semantic-conventions/diff locally for details

[codecov]

Codecov Report

❌ Patch coverage is 63.04348% with 68 lines in your changes missing coverage. Please review.
✅ Project coverage is 55.46%. Comparing base (9ff4f39) to head (891081a).

Files with missing lines	Patch %	Lines
pkg/instrumentation/shared/setup.go	35.84%	32 Missing and 2 partials ⚠️
pkg/instrumentation/grpc/semconv/grpc.go	66.66%	26 Missing ⚠️
pkg/instrumentation/grpc/semconv/util.go	81.81%	2 Missing and 2 partials ⚠️
pkg/instrumentation/grpc/semconv/semconv.go	92.00%	1 Missing and 1 partial ⚠️
tool/internal/instrument/trampoline.go	0.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff            @@
##           main     #226       +/-   ##
=========================================
+ Coverage      0   55.46%   +55.46%     
=========================================
  Files         0       45       +45     
  Lines         0     3260     +3260     
=========================================
+ Hits          0     1808     +1808     
- Misses        0     1314     +1314     
- Partials      0      138      +138     

Flag	Coverage Δ
pkg	55.46% <63.04%> (?)
tool	55.46% <63.04%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Copilot]

Pull request overview

This PR adds comprehensive gRPC instrumentation support using compile-time code injection. The implementation instruments both client (grpc.NewClient, grpc.DialContext) and server (grpc.NewServer) methods to automatically collect distributed traces and metrics without requiring code changes.

Key Changes:

• Added gRPC client and server instrumentation hooks using stats.Handler injection
• Implemented semantic conventions for gRPC following OpenTelemetry v1.37.0
• Added filtering to prevent infinite recursion with OTLP gRPC exporters
• Switched to HTTP/protobuf exporter to avoid gRPC deadlock issues

Reviewed changes

Copilot reviewed 31 out of 34 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
tool/data/grpc.yaml	Hook configuration for gRPC client/server methods
tool/internal/instrument/trampoline.go	Added ellipsis type handling for variadic parameters
pkg/instrumentation/grpc/server/	Server-side instrumentation with stats handler
pkg/instrumentation/grpc/client/	Client-side instrumentation for NewClient and DialContext
pkg/instrumentation/grpc/semconv/	Semantic conventions and utility functions
pkg/otelsetup/setup.go	Changed from gRPC to HTTP exporter to prevent deadlock
test/integration/grpc_*_test.go	Integration tests for client/server instrumentation
test/e2e/grpc_test.go	End-to-end tests
demo/grpc/	Updated demo applications with structured logging
demo/infrastructure/docker-compose/	Updated to use HTTP exporter and added gRPC Makefile targets

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.