Semantic Conventions for RPC Spans

Status: Experimental

This document defines how to describe remote procedure calls (also called “remote method invocations” / “RMI”) with spans.

Warning Existing RPC instrumentations that are using v1.20.0 of this document (or prior):

• SHOULD NOT change the version of the networking conventions that they emit until the HTTP semantic conventions are marked stable (HTTP stabilization will include stabilization of a core set of networking conventions which are also used in RPC instrumentations). Conventions include, but are not limited to, attributes, metric and span names, and unit of measure.
• SHOULD introduce an environment variable OTEL_SEMCONV_STABILITY_OPT_IN in the existing major version which is a comma-separated list of values. The only values defined so far are:
  • http - emit the new, stable networking conventions, and stop emitting the old experimental networking conventions that the instrumentation emitted previously.
  • http/dup - emit both the old and the stable networking conventions, allowing for a seamless transition.
  • The default behavior (in the absence of one of these values) is to continue emitting whatever version of the old experimental networking conventions the instrumentation was emitting previously.
  • Note: http/dup has higher precedence than http in case both values are present
• SHOULD maintain (security patching at a minimum) the existing major version for at least six months after it starts emitting both sets of conventions.
• SHOULD drop the environment variable in the next major version.

Common remote procedure call conventions

A remote procedure calls is described by two separate spans, one on the client-side and one on the server-side.

For outgoing requests, the SpanKind MUST be set to CLIENT and for incoming requests to SERVER.

Remote procedure calls can only be represented with these semantic conventions, when the names of the called service and method are known and available.

Span name

The span name MUST be the full RPC method name formatted as:

$package.$service/$method

(where $service MUST NOT contain dots and $method MUST NOT contain slashes)

If there is no package name or if it is unknown, the $package. part (including the period) is omitted.

Examples of span names:

• grpc.test.EchoService/Echo
• com.example.ExampleRmiService/exampleMethod
• MyCalcService.Calculator/Add reported by the server and MyServiceReference.ICalculator/Add reported by the client for .NET WCF calls
• MyServiceWithNoPackage/theMethod

Common attributes

[1]: The value SHOULD be normalized to lowercase.

Consider always setting the transport when setting a port number, since a port number is ambiguous without knowing the transport. For example different processes could be listening on TCP port 12345 and UDP port 12345.

[2]: The value SHOULD be normalized to lowercase.

[3]: This is the logical name of the method from the RPC interface perspective, which can be different from the name of any implementing method/function. The code.function attribute may be used to store the latter (e.g., method actually executing the call on the server side, RPC client stub method on the client side).

[4]: This is the logical name of the service from the RPC interface perspective, which can be different from the name of any implementing class. The code.namespace attribute may be used to store the latter (despite the attribute name, it may include a class name; e.g., class with method actually executing the call on the server side, RPC client stub class on the client side).

[5]: May contain server IP address, DNS name, or local socket name. When host component is an IP address, instrumentations SHOULD NOT do a reverse proxy lookup to obtain DNS name and SHOULD set server.address to the IP address provided in the host component.

[6]: When observed from the client side, and when communicating through an intermediary, server.port SHOULD represent the server port behind any intermediaries, for example proxies, if it’s available.

[7]: if the port is supported by the network transport used for communication.

network.transport has the following list of well-known values. If one of them applies, then the respective value MUST be used, otherwise a custom value MAY be used.

network.type has the following list of well-known values. If one of them applies, then the respective value MUST be used, otherwise a custom value MAY be used.

rpc.system has the following list of well-known values. If one of them applies, then the respective value MUST be used, otherwise a custom value MAY be used.

Service name

On the server process receiving and handling the remote procedure call, the service name provided in rpc.service does not necessarily have to match the service.name resource attribute. One process can expose multiple RPC endpoints and thus have multiple RPC service names. From a deployment perspective, as expressed by the service.* resource attributes, it will be treated as one deployed service with one service.name. Likewise, on clients sending RPC requests to a server, the service name provided in rpc.service does not have to match the peer.service span attribute.

As an example, given a process deployed as QuoteService, this would be the name that goes into the service.name resource attribute which applies to the entire process. This process could expose two RPC endpoints, one called CurrencyQuotes (= rpc.service) with a method called getMeanRate (= rpc.method) and the other endpoint called StockQuotes (= rpc.service) with two methods getCurrentBid and getLastClose (= rpc.method). In this example, spans representing client request should have their peer.service attribute set to QuoteService as well to match the server’s service.name resource attribute. Generally, a user SHOULD NOT set peer.service to a fully qualified RPC service name.

Client attributes

Server attributes

[1]: When observed from the server side, and when communicating through an intermediary, client.address SHOULD represent the client address behind any intermediaries, for example proxies, if it’s available.

[2]: When observed from the server side, and when communicating through an intermediary, client.port SHOULD represent the client port behind any intermediaries, for example proxies, if it’s available.

[3]: The value SHOULD be normalized to lowercase.

Consider always setting the transport when setting a port number, since a port number is ambiguous without knowing the transport. For example different processes could be listening on TCP port 12345 and UDP port 12345.

[4]: The value SHOULD be normalized to lowercase.

Events

In the lifetime of an RPC stream, an event for each message sent/received on client and server spans SHOULD be created. In case of unary calls only one sent and one received message will be recorded for both client and server spans.

The event name MUST be message.

[1]: This way we guarantee that the values will be consistent between different implementations.

message.type MUST be one of the following:

Distinction from HTTP spans

HTTP calls can generally be represented using just HTTP spans. If they address a particular remote service and method known to the caller, i.e., when it is a remote procedure call transported over HTTP, the rpc.* attributes might be added additionally on that span, or in a separate RPC span that is a parent of the transporting HTTP call. Note that method in this context is about the called remote procedure and not the HTTP verb (GET, POST, etc.).

Semantic Conventions for specific RPC technologies

More specific Semantic Conventions are defined for the following RPC technologies:

• Connect: Semantic Conventions for Connect RPC.
• gRPC: Semantic Conventions for gRPC.
• JSON-RPC: Semantic Conventions for JSON-RPC.