Trait rdkafka::producer::Producer

pub trait Producer<C = DefaultProducerContext>
where
    C: ProducerContext + 'static,
{
    // Required methods
    fn client(&self) -> &Client<C>;
    fn in_flight_count(&self) -> i32;
    fn flush<T: Into<Timeout>>(&self, timeout: T) -> KafkaResult<()>;
    fn init_transactions<T: Into<Timeout>>(&self, timeout: T) -> KafkaResult<()>;
    fn begin_transaction(&self) -> KafkaResult<()>;
    fn send_offsets_to_transaction<T: Into<Timeout>>(
        &self,
        offsets: &TopicPartitionList,
        cgm: &ConsumerGroupMetadata,
        timeout: T,
    ) -> KafkaResult<()>;
    fn commit_transaction<T: Into<Timeout>>(
        &self,
        timeout: T,
    ) -> KafkaResult<()>;
    fn abort_transaction<T: Into<Timeout>>(&self, timeout: T) -> KafkaResult<()>;

    // Provided method
    fn context(&self) -> &Arc<C> { ... }
}

Common trait for all producers.

Required Methods

fn client(&self) -> &Client<C>

Returns the Client underlying this producer.

fn in_flight_count(&self) -> i32

Returns the number of messages that are either waiting to be sent or are sent but are waiting to be acknowledged.

fn flush<T: Into<Timeout>>(&self, timeout: T) -> KafkaResult<()>

Flushes any pending messages.

This method should be called before termination to ensure delivery of all enqueued messages. It will call poll() internally.

fn init_transactions<T: Into<Timeout>>(&self, timeout: T) -> KafkaResult<()>

Enable sending transactions with this producer.

Prerequisites

• The configuration used to create the producer must include a transactional.id setting.
• You must not have sent any messages or called any of the other transaction-related functions.

Details

This function ensures any transactions initiated by previous producers with the same transactional.id are completed. Any transactions left open by any such previous producers will be aborted.

Once previous transactions have been fenced, this function acquires an internal producer ID and epoch that will be used by all transactional messages sent by this producer.

If this function returns successfully, messages may only be sent to this producer when a transaction is active. See Producer::begin_transaction.

This function may block for the specified timeout.

fn begin_transaction(&self) -> KafkaResult<()>

Begins a new transaction.

Prerequisites

You must have successfully called Producer::init_transactions.

Details

This function begins a new transaction, and implicitly associates that open transaction with this producer.

After a successful call to this function, any messages sent via this producer or any calls to Producer::send_offsets_to_transaction will be implicitly associated with this transaction, until the transaction is finished.

Finish the transaction by calling Producer::commit_transaction or Producer::abort_transaction.

While a transaction is open, you must perform at least one transaction operation every transaction.timeout.ms to avoid timing out the transaction on the broker.

fn send_offsets_to_transaction<T: Into<Timeout>>( &self, offsets: &TopicPartitionList, cgm: &ConsumerGroupMetadata, timeout: T, ) -> KafkaResult<()>

Associates an offset commit operation with this transaction.

Prerequisites

The producer must have an open transaction via a call to Producer::begin_transaction.

Details

Sends a list of topic partition offsets to the consumer group coordinator for cgm, and marks the offsets as part of the current transaction. These offsets will be considered committed only if the transaction is committed successfully.

The offsets should be the next message your application will consume, i.e., one greater than the the last processed message’s offset for each partition.

Use this method at the end of a consume-transform-produce loop, prior to comitting the transaction with Producer::commit_transaction.

This function may block for the specified timeout.

Hints

To obtain the correct consumer group metadata, call Consumer::group_metadata on the consumer for which offsets are being committed.

The consumer must not have automatic commits enabled.

fn commit_transaction<T: Into<Timeout>>(&self, timeout: T) -> KafkaResult<()>

Commits the current transaction.

Prerequisites

The producer must have an open transaction via a call to Producer::begin_transaction.

Details

Any outstanding messages will be flushed (i.e., delivered) before actually committing the transaction.

If any of the outstanding messages fail permanently, the current transaction will enter an abortable error state and this function will return an abortable error. You must then call Producer::abort_transaction before attemping to create another transaction.

This function may block for the specified timeout.

fn abort_transaction<T: Into<Timeout>>(&self, timeout: T) -> KafkaResult<()>

Aborts the current transaction.

Prerequisites

The producer must have an open transaction via a call to Producer::begin_transaction.

Details

Any oustanding messages will be purged and failed with RDKafkaErrorCode::PurgeInflight or RDKafkaErrorCode::PurgeQueue.

This function should also be used to recover from non-fatal abortable transaction errors.

This function may block for the specified timeout.

Provided Methods

fn context(&self) -> &Arc<C>

Returns a reference to the ProducerContext used to create this producer.

Object Safety

This trait is not object safe.

Implementors

impl<C> Producer<C> for BaseProducer<C>

where C: ProducerContext,

impl<C> Producer<C> for ThreadedProducer<C>

where C: ProducerContext + 'static,

impl<C, R> Producer<FutureProducerContext<C>> for FutureProducer<C, R>

where C: ClientContext + 'static, R: AsyncRuntime,