# How to write semantic conventions **Status**: [Development][DocumentStatus] - [Defining new conventions](#defining-new-conventions) - [Best practices](#best-practices) - [Prototyping](#prototyping) - [Defining attributes](#defining-attributes) - [Defining enum attribute members](#defining-enum-attribute-members) - [Defining spans](#defining-spans) - [What operation does this span represent](#what-operation-does-this-span-represent) - [Naming pattern](#naming-pattern) - [Status](#status) - [Kind](#kind) - [Attributes](#attributes) - [Defining metrics](#defining-metrics) - [Defining entities](#defining-entities) - [Defining events](#defining-events) - [Stabilizing existing conventions](#stabilizing-existing-conventions) - [Migration plan](#migration-plan) This document describes requirements, recommendations, and best practices on how to define conventions for new areas or make substantial changes to the existing ones. ## Defining new conventions - New conventions MUST have a group of codeowners. See [project management](https://github.com/open-telemetry/community/blob/main/project-management.md) for more details. - New conventions SHOULD be defined in YAML files. See [YAML Model for Semantic Conventions](/model/README.md) for the details. - New conventions SHOULD be defined with `development` stability level. - New conventions SHOULD include telemetry signal definitions (spans, metrics, events, resources, profiles) and MAY include new attribute definitions. ### Best practices > [!NOTE] > > This section contains non-normative guidance. Please read [T-Shaped Signals](t-shaped-signals.md) for guidance on how to approach creating Semantic Conventions through defining key use cases for conventions. #### Prototyping It is strongly recommended to prototype proposed conventions in one or more instrumentations and: - validate the feasibility of collecting the proposed telemetry and attributes, ensuring the information is available and can be gathered with reasonable overhead, - confirm that the proposed terminology applies across the diverse libraries and technologies covered by the conventions, - provide actionable guidance to instrumentation authors on when and how to collect attributes and record telemetry, - evaluate how the new or updated telemetry integrates with other instrumentation layers, identifying gaps, duplication, or opportunities to improve the end user experience. #### Defining attributes Reuse existing attributes when possible. Look through [existing conventions](/docs/registry/attributes/) for similar areas, check out [common attributes](/docs/general/attributes.md). Semantic conventions authors are encouraged to use attributes from different namespaces. Consider adding a new attribute if all of the following apply: - It provides a clear benefit to end users by enhancing telemetry. - There is a clear plan to use the attributes when defining spans, metrics, events, resources, or other telemetry signals in semantic conventions. - There is a clear plan on how these attributes will be used by instrumentations Semantic convention maintainers may reject the addition of a new attribute if its benefits and use-cases are not yet clear. When defining a new attribute: - Follow the [naming guidance](/docs/general/naming.md) - Provide descriptive `brief` and `note` sections to clearly explain what the attribute represents. - If the attribute represents a common concept documented externally, include relevant links. For example, always link to concepts defined in RFCs or other standards. - If the attribute's value might contain PII or other sensitive information, explicitly call this out in the `note`. Include a warning similar to the following: ```yaml - id: user.full_name ... note: | ... > [!WARNING] > > This attribute contains sensitive (PII) information. ``` - Use the appropriate [attribute type](https://github.com/open-telemetry/weaver/blob/main/schemas/semconv-syntax.md#type) - If the value has a reasonably short (open or closed) set of possible values, define it as an enum. - If the value is a timestamp, record it as a string in ISO 8601 format. - For arrays of primitives, use the array type. Avoid recording arrays as a single string. - Arrays should be homogeneous, meaning all elements share the same type and represent the same concept. For example: - Latitude and longitude should be defined as separate attributes (`geo.lat` and `geo.lon`) rather than combining them into a single array, as they represent distinct concepts. - Use the template type to define attributes with dynamic names (only the last segment of the name should be dynamic). This is useful for capturing user-defined key-value pairs, such as HTTP headers. - Represent complex values as a set of flat attributes whenever possible. - Complex or structured attributes (not listed in the [set of standard attributes](https://opentelemetry.io/docs/specs/otel/common/#attribute)) could be referenced on events and spans (![Development](https://img.shields.io/badge/-development-blue)) only. Semantic convention authors should assume that backends do not index individual properties of complex attributes, that querying or aggregating on such properties is inefficient and complicated, and that reporting complex attributes carries higher performance overhead. - Define new attributes with `development` stability. - Provide realistic examples - Attributes should generally have bounded values (for example, strings under 1 KB or small arrays). When captured data is unbounded by nature (such as queries, payloads, or raw records): - Mark the attribute in the registry with: ```yaml annotations: value: unbounded_size: true ``` This annotation renders a warning advising applications to configure SDK attribute limits and instrumentations to enforce safety size limits. - Unbounded attributes MUST NOT be `required` on any signal or attribute group. - Consider defining a corresponding reference attribute (see [attributes referencing external content](/docs/general/naming.md#attributes-referencing-external-content)). Consider the scope of the attribute and how it may evolve in the future: - When defining an attribute for a narrow use case, think about potential broader use cases. For example, if creating a system-specific attribute, evaluate whether other systems in the same domain might need a similar attribute in the future. Similarly, instead of defining a simple boolean flag indicating a success or failure, consider a more extensible approach, such as using a `foo.status_code` attribute to include additional details. - When defining a broad attribute applicable across multiple domains or systems, check for existing standards or widely accepted best practices in the industry. Avoid creating generic attributes that are not based on established standards. > [!NOTE] > > When defining conventions for areas with multiple implementations or systems — such as databases, > or cloud providers — it can take time to strike the right balance between being > overly generic and not generic enough. > > Start with experimental conventions, document how they apply to a diverse range > of providers, systems, or libraries, and prototype instrumentations. > > The end user experience should serve as the primary guiding principle: > > - If the attribute is expected to be used in general-purpose metrics for the area, > consider introducing a common attribute. > > For example, most messaging systems have a concept like a queue or topic. > Queue or topic names are critical for latency and throughput metrics and > equally important for spans to debug and visualize message flow. > This indicates the need for a generic attribute representing any type of messaging destination. > > - If the attribute represents something useful in a narrow set of scenarios or > is specific to certain system metrics, spans, or events, it likely does not need to be generic. #### Defining enum attribute members Enum attributes generally fall into three main categories: **Complete enums** document all possible values. For example, `cpu.mode` covers all known CPU modes. Metrics like `system.cpu.time` depend on having all modes defined. Authors should document all known values upfront, though new values may be added later to support new operating systems or CPU architectures. **Open enums** like `error.type`, `cloud.provider`, and `cloud.platform` allow conventions and instrumentations to define their own applicable values without adding them to semantic conventions first. Values used by only one implementation can be documented with that implementation. When proposing a well-known value here, look for another independent implementation that uses the same value and link to both in the proposal. This helps reviewers assess whether the value is useful beyond one implementation. **System identifier enums** specify a system, project, provider, product, or protocol. For example, `db.system.name` contains database names like `mongodb` or `mysql`. System identifier enums help differentiate telemetry signals. MongoDB and MySQL both follow general database conventions, but populate attributes like `db.collection.name` differently and have system-specific attributes. Each system has its own span definitions and documentation. See [MongoDB](/docs/db/mongodb.md) and [MySQL](/docs/db/mysql.md) as examples and check out [system-specific naming](../general/naming.md#system-specific-attributes) for naming guidance. > [!IMPORTANT] > System identifier enums don't need to list every possible system, component, or technology. > > OpenTelemetry instrumentations that use enum attributes SHOULD document their values before releasing **stable** artifacts and MAY support undocumented values in unstable artifacts or behind a feature flag. Only define new system identifiers when you also document how conventions apply to that system. For example, when adding a new `db.system.name` value, create documentation and span definitions that show how generic attributes work for that database. #### Defining spans Spans describe individual executions of specific operations within a trace. When to define spans: - The corresponding operation is significant for your observability needs. - The operation has duration. For example, define spans for operations that involve one or more network calls. > [!NOTE] > > Known exception: [messaging `create`](/docs/messaging/messaging-spans.md#operation-types) span > is defined for a local call. This is necessary when publishing batches of > messages to ensure each message has a unique context and can be traced > individually end-to-end. When not to define spans: - For point-in-time occurrences - use events instead. - For short operations that don't involve out-of-process calls, such as serialization or deserialization. - If there is an existing span definition that captures a very similar operation. For example, a DB client span represents DB query execution from ORM or DB driver perspectives. Both layers could be instrumented, but inner layers may be suppressed to reduce duplication. > [!IMPORTANT] > > It's a common practice to accompany a span definition with: > > - A metric that measures the duration of the same operation. > - An event that records exceptions that prevent this operation from completing > successfully. > > For example, the `http.client.request.duration` metric and > `http.client.request.exception` event are recorded alongside the corresponding > HTTP client span. A span definition should describe the [operation it represents](#what-operation-does-this-span-represent), the [naming pattern](#naming-pattern), considerations for setting span [status](#status), the [span kind](#kind), and the list of applicable [attributes](#attributes). ##### What operation does this span represent Define the scope and boundaries of the operation: - When the span starts and ends. - If this span represents a client call, specify whether it captures the logical call (as observed by the API caller) or the physical call (per-attempt). - Define a different span for different operations - e.g., when spans have different kinds or a significantly different set of attributes. For example, HTTP client and server spans are two independent definitions. Messaging publishing and receiving are also different span types. ##### Naming pattern - Span names must have low cardinality and should provide a reasonable grouping for that operation. See [Span name guidelines](https://opentelemetry.io/docs/specs/otel/trace/api/#span) for details. - Span names usually follow the `{action} {target}` pattern. For example, `send orders_queue`. - Span names should only include information that's available as span attributes. That is, `{action}` and `{target}` are usually also available as attributes and are used on metrics describing that operation. - Static text should not be included in span names but can be used as a fallback. For example, we use `GET /orders/{id}` instead of `HTTP GET /orders/{id}` for HTTP server span names. - Provide fallback values in case some of the attributes used in the span name are not available or could be problematic in edge cases (e.g., have high cardinality). - If a span name can become too long, define limits and truncation strategies (e.g., DB conventions define a 255-character limit). ##### Status Define what constitutes an error for that operation. If there are no special considerations, reference the [Recording errors](/docs/general/recording-errors.md) document. Certain conditions can't be clearly classified as errors or non-errors (such as cancellations, HTTP 404, and others). Avoid using strict requirements — allow instrumentations to leverage additional context to provide a more accurate status. ##### Kind All span definitions MUST include a specific [span kind](https://opentelemetry.io/docs/specs/otel/trace/api/#spankind). One span definition can only mention one span kind. ##### Attributes Capture only the important details of the specific operation. Parent operations or sub-operations will have their own spans. For example, when recording a call to upload a file to an object store, include the endpoint, operation name (such as upload file), collection, and object identifier. Don't include details of the underlying HTTP/gRPC requests unless there is a strong reason to do so. Only include attributes that bring clear value — this keeps telemetry volume and performance overhead low. Don't try to capture all available details. When in doubt, don't reference additional attributes - they can be added incrementally based on feedback. Define which additional properties this span needs to be useful: - Include the `error.type` attribute. If the operation you're describing typically has a domain-specific error code, include that as a separate attribute as well. Document which error codes constitute an error. - Include `server.address` and `server.port` on client spans. They represent the logical server name and port; semantic conventions that refer to these attributes SHOULD specify what they mean in their context. - Include applicable `network.*` attributes on spans that describe network calls. See [Choosing network address and port attributes](/docs/network/choosing-network-attributes.md) for help picking among `client.*` / `server.*`, `source.*` / `destination.*`, and `network.local.*` / `network.peer.*`. - Include some form of operation name to describe the action being performed. For example, in the case of HTTP, it's `http.request.method`; in the case of RPC, it's `rpc.method`; for messaging, `messaging.operation.name`; and for GenAI, `gen_ai.operation.name`. This attribute typically serves as the `{action}` in the span name and may be used across multiple span definitions within the same domain. - Identify other important characteristics such as the operation target (DB collection, messaging queue, GenAI model, object store collection), input parameters, and result properties that should be recorded on the span. - When referencing an attribute: - Specify if an attribute is relevant for head-sampling. Such attributes should be provided at start time so that they will be passed to the sampler. Usually, these are attributes that have low cardinality and are easy to obtain. - Specify [requirement level](/docs/general/attribute-requirement-level.md). Only absolutely essential (and always available) attributes can be `required`. Attributes that may include sensitive information, are expensive to obtain, or are verbose, should be `opt-in`. - Update the brief and note to tailor the attribute definition to that operation. #### Defining metrics TBD #### Defining entities Follow the [Entity Modeling Guide](resource-and-entities.md). #### Defining events Follow the [Semantic conventions for events](/docs/general/events.md). ## Stabilizing existing conventions - All conventions MUST be defined in YAML before they can be declared stable - Conventions that are not used by instrumentations MUST NOT be declared stable TODO: - migration plan ### Migration plan TODO [DocumentStatus]: https://opentelemetry.io/docs/specs/otel/document-status/