[][src]Crate tracing

A scoped, structured logging and diagnostics system.

Overview

tracing is a framework for instrumenting Rust programs to collect structured, event-based diagnostic information.

In asynchronous systems like Tokio, interpreting traditional log messages can often be quite challenging. Since individual tasks are multiplexed on the same thread, associated events and log lines are intermixed making it difficult to trace the logic flow. tracing expands upon logging-style diagnostics by allowing libraries and applications to record structured events with additional information about temporality and causality — unlike a log message, a span in tracing has a beginning and end time, may be entered and exited by the flow of execution, and may exist within a nested tree of similar spans. In addition, tracing spans are structured, with the ability to record typed data as well as textual messages.

The tracing crate provides the APIs necessary for instrumenting libraries and applications to emit trace data.

Core Concepts

The core of tracing's API is composed of spans, events and subscribers. We'll cover these in turn.

Spans

To record the flow of execution through a program, tracing introduces the concept of spans. Unlike a log line that represents a moment in time, a span represents a period of time with a beginning and an end. When a program begins executing in a context or performing a unit of work, it enters that context's span, and when it stops executing in that context, it exits the span. The span in which a thread is currently executing is referred to as that thread's current span.

For example:

use tracing::{span, Level};
let span = span!(Level::TRACE, "my_span");
// `enter` returns a RAII guard which, when dropped, exits the span. this
// indicates that we are in the span for the current lexical scope.
let _enter = span.enter();
// perform some work in the context of `my_span`...

The span module's documentation provides further details on how to use spans.

⚠ ️Warning
     Warning: In asynchronous code that uses async/await syntax,
     Span::enter may produce incorrect traces if the returned drop
     guard is held across an await point. See
     the method documentation
     for details.
 

Events

An Event represents a moment in time. It signifies something that happened while a trace was being recorded. Events are comparable to the log records emitted by unstructured logging code, but unlike a typical log line,``s an Event may occur within the context of a span.

For example:

use tracing::{event, span, Level};

// records an event outside of any span context:
event!(Level::INFO, "something happened");

let span = span!(Level::INFO, "my_span");
let _guard = span.enter();

// records an event within "my_span".
event!(Level::DEBUG, "something happened inside my_span");

In general, events should be used to represent points in time within a span — a request returned with a given status code, n new items were taken from a queue, and so on.

The Event struct documentation provides further details on using events.

Subscribers

As Spans and Events occur, they are recorded or aggregated by implementations of the Subscriber trait. Subscribers are notified when an Event takes place and when a Span is entered or exited. These notifications are represented by the following Subscriber trait methods:

  • event, called when an Event takes place,
  • enter, called when execution enters a Span,
  • exit, called when execution exits a Span

In addition, subscribers may implement the enabled function to filter the notifications they receive based on metadata describing each Span or Event. If a call to Subscriber::enabled returns false for a given set of metadata, that Subscriber will not be notified about the corresponding Span or Event. For performance reasons, if no currently active subscribers express interest in a given set of metadata by returning true, then the corresponding Span or Event will never be constructed.

Usage

First, add this to your Cargo.toml:

[dependencies]
tracing = "0.1"

Compiler support: requires rustc 1.39+

Recording Spans and Events

Spans and events are recorded using macros.

Spans

The span! macro expands to a Span struct which is used to record a span. The Span::enter method on that struct records that the span has been entered, and returns a RAII guard object, which will exit the span when dropped.

For example:

use tracing::{span, Level};
// Construct a new span named "my span" with trace log level.
let span = span!(Level::TRACE, "my span");

// Enter the span, returning a guard object.
let _enter = span.enter();

// Any trace events that occur before the guard is dropped will occur
// within the span.

// Dropping the guard will exit the span.

The #[instrument] attribute provides an easy way to add tracing spans to functions. A function annotated with #[instrument] will create and enter a span with that function's name every time the function is called, with arguments to that function will be recorded as fields using fmt::Debug.

For example:

This example is not tested
use tracing::{Level, event, instrument};

#[instrument]
pub fn my_function(my_arg: usize) {
    // This event will be recorded inside a span named `my_function` with the
    // field `my_arg`.
    event!(Level::INFO, "inside my_function!");
    // ...
}
Note
     Note: Using #[instrument] on
     async fns requires the 
     tracing-futures crate as a dependency as well.
 

You can find more examples showing how to use this crate here.

Events

Events are recorded using the event! macro:

use tracing::{event, Level};
event!(Level::INFO, "something has happened!");

Using the Macros

The span! and event! macros use fairly similar syntax, with some exceptions.

Configuring Attributes

Both macros require a Level specifying the verbosity of the span or event. Optionally, the target and parent span may be overridden. If the target and parent span are not overridden, they will default to the module path where the macro was invoked and the current span (as determined by the subscriber), respectively.

For example:

span!(target: "app_spans", Level::TRACE, "my span");
event!(target: "app_events", Level::INFO, "something has happened!");
let span = span!(Level::TRACE, "my span");
event!(parent: &span, Level::INFO, "something has happened!");

The span macros also take a string literal after the level, to set the name of the span.

Recording Fields

Structured fields on spans and events are specified using the syntax field_name = field_value. Fields are separated by commas.

// records an event with two fields:
//  - "answer", with the value 42
//  - "question", with the value "life, the universe and everything"
event!(Level::INFO, answer = 42, question = "life, the universe, and everything");

As shorthand, local variables may be used as field values without an assignment, similar to struct initializers. For example:

let user = "ferris";

span!(Level::TRACE, "login", user);
// is equivalent to:
span!(Level::TRACE, "login", user = user);

Field names can include dots, but should not be terminated by them:

let user = "ferris";
let email = "ferris@rust-lang.org";
span!(Level::TRACE, "login", user, user.email = email);

Since field names can include dots, fields on local structs can be used using the local variable shorthand:

let user = User {
    name: "ferris",
    email: "ferris@rust-lang.org",
};
// the span will have the fields `user.name = "ferris"` and
// `user.email = "ferris@rust-lang.org"`.
span!(Level::TRACE, "login", user.name, user.email);

The ? sigil is shorthand that specifies a field should be recorded using its fmt::Debug implementation:

#[derive(Debug)]
struct MyStruct {
    field: &'static str,
}

let my_struct = MyStruct {
    field: "Hello world!"
};

// `my_struct` will be recorded using its `fmt::Debug` implementation.
event!(Level::TRACE, greeting = ?my_struct);
// is equivalent to:
event!(Level::TRACE, greeting = tracing::field::debug(&my_struct));

The % sigil operates similarly, but indicates that the value should be recorded using its fmt::Display implementation:

// `my_struct.field` will be recorded using its `fmt::Display` implementation.
event!(Level::TRACE, greeting = %my_struct.field);
// is equivalent to:
event!(Level::TRACE, greeting = tracing::field::display(&my_struct.field));

The % and ? sigils may also be used with local variable shorthand:

// `my_struct.field` will be recorded using its `fmt::Display` implementation.
event!(Level::TRACE, %my_struct.field);

Additionally, a span may declare fields with the special value Empty, which indicates that that the value for that field does not currently exist but may be recorded later. For example:

use tracing::{trace_span, field};

// Create a span with two fields: `greeting`, with the value "hello world", and
// `parting`, without a value.
let span = trace_span!("my_span", greeting = "hello world", parting = field::Empty);

// ...

// Now, record a value for parting as well.
span.record("parting", &"goodbye world!");

Note that a span may have up to 32 fields. The following will not compile:

This example deliberately fails to compile
let bad_span = span!(
    Level::TRACE,
    "too many fields!",
    a = 1, b = 2, c = 3, d = 4, e = 5, f = 6, g = 7, h = 8, i = 9,
    j = 10, k = 11, l = 12, m = 13, n = 14, o = 15, p = 16, q = 17,
    r = 18, s = 19, t = 20, u = 21, v = 22, w = 23, x = 24, y = 25,
    z = 26, aa = 27, bb = 28, cc = 29, dd = 30, ee = 31, ff = 32, gg = 33
);

Finally, events may also include human-readable messages, in the form of a format string and (optional) arguments, after the event's key-value fields. If a format string and arguments are provided, they will implicitly create a new field named message whose value is the provided set of format arguments.

For example:

let question = "the answer to the ultimate question of life, the universe, and everything";
let answer = 42;
// records an event with the following fields:
// - `question.answer` with the value 42,
// - `question.tricky` with the value `true`,
// - "message", with the value "the answer to the ultimate question of life, the
//    universe, and everything is 42."
event!(
    Level::DEBUG,
    question.answer = answer,
    question.tricky = true,
    "the answer to {} is {}.", question, answer
);

Specifying a formatted message in this manner does not allocate by default.

Shorthand Macros

tracing also offers a number of macros with preset verbosity levels. The trace!, debug!, info!, warn!, and error! behave similarly to the event! macro, but with the Level argument already specified, while the corresponding trace_span!, debug_span!, info_span!, warn_span!, and error_span! macros are the same, but for the span! macro.

These are intended both as a shorthand, and for compatibility with the log crate (see the next section).

For log Users

Users of the log crate should note that tracing exposes a set of macros for creating Events (trace!, debug!, info!, warn!, and error!) which may be invoked with the same syntax as the similarly-named macros from the log crate. Often, the process of converting a project to use tracing can begin with a simple drop-in replacement.

Let's consider the log crate's yak-shaving example:

This example is not tested
use std::{error::Error, io};
use tracing::{debug, error, info, span, warn, Level};

// the `#[tracing::instrument]` attribute creates and enters a span
// every time the instrumented function is called. The span is named after the
// the function or method. Parameters passed to the function are recorded as fields.
#[tracing::instrument]
pub fn shave(yak: usize) -> Result<(), Box<dyn Error + 'static>> {
    // this creates an event at the DEBUG level with two fields:
    // - `excitement`, with the key "excitement" and the value "yay!"
    // - `message`, with the key "message" and the value "hello! I'm gonna shave a yak."
    //
    // unlike other fields, `message`'s shorthand initialization is just the string itself.
    debug!(excitement = "yay!", "hello! I'm gonna shave a yak.");
    if yak == 3 {
        warn!("could not locate yak!");
        // note that this is intended to demonstrate `tracing`'s features, not idiomatic
        // error handling! in a library or application, you should consider returning
        // a dedicated `YakError`. libraries like snafu or thiserror make this easy.
        return Err(io::Error::new(io::ErrorKind::Other, "shaving yak failed!").into());
    } else {
        debug!("yak shaved successfully");
    }
    Ok(())
}

pub fn shave_all(yaks: usize) -> usize {
    // Constructs a new span named "shaving_yaks" at the TRACE level,
    // and a field whose key is "yaks". This is equivalent to writing:
    //
    // let span = span!(Level::TRACE, "shaving_yaks", yaks = yaks);
    //
    // local variables (`yaks`) can be used as field values
    // without an assignment, similar to struct initializers.
    let span = span!(Level::TRACE, "shaving_yaks", yaks);
    let _enter = span.enter();

    info!("shaving yaks");

    let mut yaks_shaved = 0;
    for yak in 1..=yaks {
        let res = shave(yak);
        debug!(yak, shaved = res.is_ok());

        if let Err(ref error) = res {
            // Like spans, events can also use the field initialization shorthand.
            // In this instance, `yak` is the field being initalized.
            error!(yak, error = error.as_ref(), "failed to shave yak!");
        } else {
            yaks_shaved += 1;
        }
        debug!(yaks_shaved);
    }

    yaks_shaved
}

In libraries

Libraries should link only to the tracing crate, and use the provided macros to record whatever information will be useful to downstream consumers.

In executables

In order to record trace events, executables have to use a Subscriber implementation compatible with tracing. A Subscriber implements a way of collecting trace data, such as by logging it to standard output.

This library does not contain any Subscriber implementations; these are provided by other crates.

The simplest way to use a subscriber is to call the set_global_default function:

extern crate tracing;

let my_subscriber = FooSubscriber::new();
tracing::subscriber::set_global_default(my_subscriber)
    .expect("setting tracing default failed");
⚠ ️Warning
 Warning: In general, libraries should not call
 set_global_default()! Doing so will cause conflicts when
 executables that depend on the library try to set the default later.
 

This subscriber will be used as the default in all threads for the remainder of the duration of the program, similar to setting the logger in the log crate.

In addition, the default subscriber can be set through using the with_default function. This follows the tokio pattern of using closures to represent executing code in a context that is exited at the end of the closure. For example:


let my_subscriber = FooSubscriber::new();
tracing::subscriber::with_default(my_subscriber, || {
    // Any trace events generated in this closure or by functions it calls
    // will be collected by `my_subscriber`.
})

This approach allows trace data to be collected by multiple subscribers within different contexts in the program. Note that the override only applies to the currently executing thread; other threads will not see the change from with_default.

Any trace events generated outside the context of a subscriber will not be collected.

Once a subscriber has been set, instrumentation points may be added to the executable using the tracing crate's macros.

log Compatibility

The log crate provides a simple, lightweight logging facade for Rust. While tracing builds upon log's foundation with richer structured diagnostic data, log's simplicity and ubiquity make it the "lowest common denominator" for text-based logging in Rust — a vast majority of Rust libraries and applications either emit or consume log records. Therefore, tracing provides multiple forms of interoperability with log: tracing instrumentation can emit log records, and a compatibility layer enables tracing Subscribers to consume log records as tracing Events.

Emitting log Records

This crate provides two feature flags, "log" and "log-always", which will cause [spans] and [events] to emit log records. When the "log" feature is enabled, if no tracing Subscriber is active, invoking an event macro or creating a span with fields will emit a log record. This is intended primarily for use in libraries which wish to emit diagnostics that can be consumed by applications using tracing or log, without paying the additional overhead of emitting both forms of diagnostics when tracing is in use.

Enabling the "log-always" feature will cause log records to be emitted even if a tracing Subscriber is set. This is intended to be used in applications where a log Logger is being used to record a textual log, and tracing is used only to record other forms of diagnostics (such as metrics, profiling, or distributed tracing data). Unlike the "log" feature, libraries generally should not enable the "log-always" feature, as doing so will prevent applications from being able to opt out of the log records.

See here for more details on this crate's feature flags.

The generated log records' messages will be a string representation of the span or event's fields, and all additional information recorded by log (target, verbosity level, module path, file, and line number) will also be populated. Additionally, log records are also generated when spans are entered, exited, and closed. Since these additional span lifecycle logs have the potential to be very verbose, and don't include additional fields, they are categorized under a separate log target, "tracing::span", which may be enabled or disabled separately from other log records emitted by tracing.

Consuming log Records

The tracing-log crate provides a compatibility layer which allows a tracing Subscriber to consume log records as though they were tracing [events]. This allows applications using tracing to record the logs emitted by dependencies using log as events within the context of the application's trace tree. See that crate's documentation for details.

In addition to tracing and tracing-core, the tokio-rs/tracing repository contains several additional crates designed to be used with the tracing ecosystem. This includes a collection of Subscriber implementations, as well as utility and adapter crates to assist in writing Subscribers and instrumenting applications.

In particular, the following crates are likely to be of interest:

  • tracing-futures provides a compatibility layer with the futures crate, allowing spans to be attached to Futures, Streams, and Executors.
  • tracing-subscriber provides Subscriber implementations and utilities for working with Subscribers. This includes a FmtSubscriber FmtSubscriber for logging formatted trace data to stdout, with similar filtering and formatting to the env_logger crate.
  • tracing-log provides a compatibility layer with the log crate, allowing log messages to be recorded as tracing Events within the trace tree. This is useful when a project using tracing have dependencies which use log. Note that if you're using tracing-subscriber's FmtSubscriber, you don't need to depend on tracing-log directly.
  • tracing-appender provides utilities for outputting tracing data, including a file appender and non blocking writer.

Additionally, there are also several third-party crates which are not maintained by the tokio project. These include:

  • tracing-timing implements inter-event timing metrics on top of tracing. It provides a subscriber that records the time elapsed between pairs of tracing events and generates histograms.
  • tracing-opentelemetry provides a subscriber for emitting traces to OpenTelemetry-compatible distributed tracing systems.
  • tracing-honeycomb Provides a layer that reports traces spanning multiple machines to honeycomb.io. Backed by tracing-distributed.
  • tracing-distributed Provides a generic implementation of a layer that reports traces spanning multiple machines to some backend.
  • tracing-actix provides tracing integration for the actix actor framework.
  • tracing-gelf implements a subscriber for exporting traces in Greylog GELF format.
  • tracing-coz provides integration with the coz causal profiler (Linux-only).
  • tracing-bunyan-formatter provides a layer implementation that reports events and spans in bunyan format, enriched with timing information.

If you're the maintainer of a tracing ecosystem crate not listed above, please let us know! We'd love to add your project to the list!

Note
 Note: Some of these ecosystem crates are currently
 unreleased and/or in earlier stages of development. They may be less stable
 than tracing and tracing-core.
 

Crate Feature Flags

The following crate feature flags are available:

  • A set of features controlling the static verbosity level.

  • log: causes trace instrumentation points to emit log records as well as trace events, if a default tracing subscriber has not been set. This is intended for use in libraries whose users may be using either tracing or log.

  • log-always: Emit log records from all tracing spans and events, even a tracing subscriber has been set. This should be set only by applications which intend to collect traces and logs separately; if an adapter is used to convert log records into tracing events, this will cause duplicate events to occur.

  • attributes: Includes support for the #[instrument] attribute. This is on by default, but does bring in the syn crate as a dependency, which may add to the compile time of crates that do not already use it.

  • std: Depend on the Rust standard library (enabled by default).

    no_std users may disable this feature with default-features = false:

    [dependencies]
    tracing = { version = "0.1.16", default-features = false }
    

    Compiler support: requires rustc 1.39+

Note
 Note: tracing's no_std support
 requires liballoc.
 

Modules

dispatcher

Dispatches trace events to Subscribers.

event

Events represent single points in time during the execution of a program.

field

Structured data associated with Spans and Events.

level_filters

Trace verbosity level filtering.

span

Spans represent periods of time in which a program was executing in a particular context.

subscriber

Collects and records trace data.

Macros

debug

Constructs an event at the debug level.

debug_span

Constructs a span at the debug level.

error

Constructs an event at the error level.

error_span

Constructs a span at the error level.

event

Constructs a new Event.

info

Constructs an event at the info level.

info_span

Constructs a span at the info level.

span

Constructs a new span.

trace

Constructs an event at the trace level.

trace_span

Constructs a span at the trace level.

warn

Constructs an event at the warn level.

warn_span

Constructs a span at the warn level.

Structs

Dispatch

Dispatch trace data to a Subscriber.

Event

Events represent single points in time where something occurred during the execution of a program.

Level

Describes the level of verbosity of a span or event.

Metadata

Metadata describing a span or event.

Span

A handle representing a span, with the capability to enter the span if it exists.

Traits

Subscriber

Trait representing the functions required to collect trace data.

Value

A field value of an erased type.

Attribute Macros

instrumentfeature="attributes"

Instruments a function to create and enter a tracing span every time the function is called.