Struct tracing_fluentd::Builder

pub struct Builder<F = NestedFmt, A = fn() -> Result<TcpStream>> { /* private fields */ }

Builder to enable forwarding tracing events towards the fluentd server.

Type params

• F - Attributes formatter, determines how to compose fluent::Record.
• A - function that returns Fluentd wrter. Default is to create tcp socket towards 127.0.0.1:24224 with timeout of 1s.

Implementations

impl Builder

pub fn new(tag: &'static str) -> Self

Creates default configuration.

Params:

tag - Event category to send for each record.

impl<A: MakeWriter> Builder<NestedFmt, A>

pub fn flatten(self) -> Builder<FlattenFmt, A>

Configures to flatten span/metadata attributes within record. Instead of the default nesting behavior.

impl<F: FieldFormatter, A: MakeWriter> Builder<F, A>

pub fn with_formatter<NF: FieldFormatter>(self, fmt: NF) -> Builder<NF, A>

Provides callback to get writer where to write records.

Normally fluentd server expects connection to be closed immediately upon sending records. hence created writer is dropped immediately upon writing being finished.

pub fn with_writer<MW: MakeWriter>(self, writer: MW) -> Builder<F, MW>

Provides callback to get writer where to write records.

Normally fluentd server expects connection to be closed immediately upon sending records. hence created writer is dropped immediately upon writing being finished.

pub fn layer(self) -> Result<Layer<F, ThreadWorker>, Error>

Creates tracing layer.

If you do not want to create multiple threads, consider using layer_guarded/layer_from_guard.

Error can happen during creation of worker thread.

pub fn layer_guarded(
    self
) -> Result<(Layer<F, WorkerChannel>, FlushingGuard), Error>

Creates tracing layer, returning guard that allows to stop fluentd worker on Drop.

This may be necessary due to bad API that tracing provides to control lifetime of global logger. As underlying implementations employs caching, it needs to perform flush once logger is no longer necessary hence this API is provided.

Error can happen during creation of worker thread.

pub fn layer_from_guard(self, guard: &FlushingGuard) -> Layer<F, WorkerChannel>

Creates tracing layer, using guard returned by layer_guarded.

Specifically, it will use the same worker thread as first instance of layer_guarded, without affecting lifetime of guard. Hence once guard is dropped, worker for all connected layers will stop sending logs.

Error can happen during creation of worker thread.