kafka

package module
v0.1.0 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Jul 16, 2018 License: MIT Imports: 24 Imported by: 0

README

kafka-go CircleCI Go Report Card GoDoc

Motivations

We rely on both Go and Kafka a lot at Segment. Unfortunately, the state of the Go client libraries for Kafka at the time of this writing was not ideal. The available options were:

  • sarama, which is by far the most popular but is quite difficult to work with. It is poorly documented, the API exposes low level concepts of the Kafka protocol, and it doesn't support recent Go features like contexts. It also passes all values as pointers which causes large numbers of dynamic memory allocations, more frequent garbage collections, and higher memory usage.

  • confluent-kafka-go is a cgo based wrapper around librdkafka, which means it introduces a dependency to a C library on all Go code that uses the package. It has much better documentation than sarama but still lacks support for Go contexts.

  • goka is a more recent Kafka client for Go which focuses on a specific usage pattern. It provides abstractions for using Kafka as a message passing bus between services rather than an ordered log of events, but this is not the typical use case of Kafka for us at Segment. The package also depends on sarama for all interactions with Kafka.

This is where kafka-go comes into play. It provides both low and high level APIs for interacting with Kafka, mirroring concepts and implementing interfaces of the Go standard library to make it easy to use and integrate with existing software.

Connection GoDoc

The Conn type is the core of the kafka-go package. It wraps around a raw network connection to expose a low-level API to a Kafka server.

Here are some examples showing typical use of a connection object:

// to produce messages
topic := "my-topic"
partition := 0

conn, _ := kafka.DialLeader(context.Background(), "tcp", "localhost:9092", topic, partition)

conn.SetWriteDeadline(time.Now().Add(10*time.Second))
conn.WriteMessages(
    kafka.Message{Value: []byte("one!")},
    kafka.Message{Value: []byte("two!")},
    kafka.Message{Value: []byte("three!")},
)

conn.Close()

// to consume messages
topic := "my-topic"
partition := 0

conn, _ := kafka.DialLeader(context.Background(), "tcp", "localhost:9092", topic, partition)

conn.SetReadDeadline(time.Now().Add(10*time.Second))
batch := conn.ReadBatch(10e3, 1e6) // fetch 10KB min, 1MB max

b := make([]byte, 10e3) // 10KB max per message
for {
    _, err := batch.Read(b)
    if err != nil {
        break
    }
    fmt.Println(string(b))
}

batch.Close()
conn.Close()

Because it is low level, the Conn type turns out to be a great building block for higher level abstractions, like the Reader for example.

Reader GoDoc

A Reader is another concept exposed by the kafka-go package, which intends to make it simpler to implement the typical use case of consuming from a single topic-partition pair. A Reader also automatically handles reconnections and offset management, and exposes an API that supports asynchronous cancellations and timeouts using Go contexts.

// make a new reader that consumes from topic-A, partition 0, at offset 42
r := kafka.NewReader(kafka.ReaderConfig{
    Brokers:   []string{"localhost:9092"},
    Topic:     "topic-A",
    Partition: 0,
    MinBytes:  10e3, // 10KB
    MaxBytes:  10e6, // 10MB
})
r.SetOffset(42)

for {
    m, err := r.ReadMessage(context.Background())
    if err != nil {
        break
    }
    fmt.Printf("message at offset %d: %s = %s\n", m.Offset, string(m.Key), string(m.Value))
}

r.Close()
Consumer Groups

kafka-go also supports Kafka consumer groups including broker managed offsets. To enable consumer groups, simplify specify the GroupID in the ReaderConfig.

ReadMessage automatically commits offsets when using consumer groups.

// make a new reader that consumes from topic-A
r := kafka.NewReader(kafka.ReaderConfig{
    Brokers:   []string{"localhost:9092"},
    GroupID:   "consumer-group-id",
    Topic:     "topic-A",
    MinBytes:  10e3, // 10KB
    MaxBytes:  10e6, // 10MB
})

for {
    m, err := r.ReadMessage(context.Background())
    if err != nil {
        break
    }
    fmt.Printf("message at topic/partition/offset %v/%v/%v: %s = %s\n", m.Topic, m.Partition, m.Offset, string(m.Key), string(m.Value))
}

r.Close()

There are a number of limitations when using consumer groups:

  • (*Reader).SetOffset will return an error when GroupID is set
  • (*Reader).Offset will always return -1 when GroupID is set
  • (*Reader).Lag will always return -1 when GroupID is set
  • (*Reader).ReadLag will return an error when GroupID is set
  • (*Reader).Stats will return a partition of -1 when GroupID is set
Explicit Commits

kafka-go also supports explicit commits. Instead of calling ReadMessage, call FetchMessage followed by CommitMessages.

ctx := context.Background()
for {
    m, err := r.FetchMessage(ctx)
    if err != nil {
        break
    }
    fmt.Printf("message at topic/partition/offset %v/%v/%v: %s = %s\n", m.Topic, m.Partition, m.Offset, string(m.Key), string(m.Value))
    r.CommitMessages(ctx, m)
}
Managing Commits

By default, CommitMessages will synchronously commit offsets to Kafka. For improved performance, you can instead periodically commit offsets to Kafka by setting CommitInterval on the ReaderConfig.

// make a new reader that consumes from topic-A
r := kafka.NewReader(kafka.ReaderConfig{
    Brokers:        []string{"localhost:9092"},
    GroupID:        "consumer-group-id",
    Topic:          "topic-A",
    MinBytes:       10e3, // 10KB
    MaxBytes:       10e6, // 10MB
    CommitInterval: time.Second, // flushes commits to Kafka every second
})

Writer GoDoc

To produce messages to Kafka, a program may use the low-level Conn API, but the package also provides a higher level Writer type which is more appropriate to use in most cases as it provides additional features:

  • Automatic retries and reconnections on errors.
  • Configurable distribution of messages across available partitions.
  • Synchronous or asynchronous writes of messages to Kafka.
  • Asynchronous cancellation using contexts.
  • Flushing of pending messages on close to support graceful shutdowns.
// make a writer that produces to topic-A, using the least-bytes distribution
w := kafka.NewWriter(kafka.WriterConfig{
	Brokers: []string{"localhost:9092"},
	Topic:   "topic-A",
	Balancer: &kafka.LeastBytes{},
})

w.WriteMessages(context.Background(),
	kafka.Message{
		Key:   []byte("Key-A"),
		Value: []byte("Hello World!"),
	},
	kafka.Message{
		Key:   []byte("Key-B"),
		Value: []byte("One!"),
	},
	kafka.Message{
		Key:   []byte("Key-C"),
		Value: []byte("Two!"),
	},
)

w.Close()

Note: Even though kafka.Message contain Topic and Partition fields, they MUST NOT be set when writing messages. They are intended for read use only.

Compatibility with Sarama

If you're switching from Sarama and need/want to use the same algorithm for message partitioning, you can use the kafka.Hash balancer. kafka.Hash routes messages to the same partitions that sarama's default partitioner would route to.

w := kafka.NewWriter(kafka.WriterConfig{
	Brokers: []string{"localhost:9092"},
	Topic:   "topic-A",
	Balancer: &kafka.Hash{},
})
Compression

Compression can be enable on the writer :

w := kafka.NewWriter(kafka.WriterConfig{
	Brokers: []string{"localhost:9092"},
	Topic:   "topic-A",
	CompressionCodec: snappy.NewCompressionCodec(),
})

The reader will by default figure out if the consumed messages are compressed by intepreting the message attributes.

TLS Support

For a bare bones Conn type or in the Reader/Writer configs you can specify a dialer option for TLS support. If the TLS field is nil, it will not connect with TLS.

Connection
dialer := &kafka.Dialer{
    Timeout:   10 * time.Second,
    DualStack: true,
    TLS:       &tls.Config{...tls config...},
}

conn, err := dialer.DialContext(ctx, "tcp", "localhost:9093")
Reader
dialer := &kafka.Dialer{
    Timeout:   10 * time.Second,
    DualStack: true,
    TLS:       &tls.Config{...tls config...},
}

r := kafka.NewReader(kafka.ReaderConfig{
    Brokers:        []string{"localhost:9093"},
    GroupID:        "consumer-group-id",
    Topic:          "topic-A",
    Dialer:         dialer,
})
Writer
dialer := &kafka.Dialer{
    Timeout:   10 * time.Second,
    DualStack: true,
    TLS:       &tls.Config{...tls config...},
}

w := kafka.NewWriter(kafka.WriterConfig{
	Brokers: []string{"localhost:9093"},
	Topic:   "topic-A",
	Balancer: &kafka.Hash{},
	Dialer:   dialer,
})

Documentation

Index

Examples

Constants

View Source 
const (
	SeekStart    = 0 // Seek relative to the first offset available in the partition.
	SeekAbsolute = 1 // Seek to an absolute offset.
	SeekEnd      = 2 // Seek relative to the last offset available in the partition.
	SeekCurrent  = 3 // Seek relative to the current offset.
)
View Source 
const CompressionNoneCode = 0
View Source 
const DefaultCompressionLevel int = -1

Variables

View Source 
var (
	// DefaultClientID is the default value used as ClientID of kafka
	// connections.
	DefaultClientID string
)
View Source 
var DefaultDialer = &Dialer{
	Timeout:   10 * time.Second,
	DualStack: true,
}

DefaultDialer is the default dialer used when none is specified.

Functions

func RegisterCompressionCodec 

func RegisterCompressionCodec(codec func() CompressionCodec)

RegisterCompressionCodec registers a compression codec so it can be used by a Writer.

Types

type Balancer 

type Balancer interface {
	// Balance receives a message and a set of available partitions and
	// returns the partition number that the message should be routed to.
	//
	// An application should refrain from using a balancer to manage multiple
	// sets of partitions (from different topics for examples), use one balancer
	// instance for each partition set, so the balancer can detect when the
	// partitions change and assume that the kafka topic has been rebalanced.
	Balance(msg Message, partitions ...int) (partition int)
}

The Balancer interface provides an abstraction of the message distribution logic used by Writer instances to route messages to the partitions available on a kafka cluster.

Instances of Balancer do not have to be safe to use concurrently by multiple goroutines, the Writer implementation ensures that calls to Balance are synchronized.

type BalancerFunc 

type BalancerFunc func(Message, ...int) int

BalancerFunc is an implementation of the Balancer interface that makes it possible to use regular functions to distribute messages across partitions.

func (BalancerFunc) Balance 

func (f BalancerFunc) Balance(msg Message, partitions ...int) int

Balance calls f, satisfies the Balancer interface.

type Batch 

type Batch struct {
	// contains filtered or unexported fields
}

A Batch is an iterator over a sequence of messages fetched from a kafka server.

Batches are created by calling (*Conn).ReadBatch. They hold a internal lock on the connection, which is released when the batch is closed. Failing to call a batch's Close method will likely result in a dead-lock when trying to use the connection.

Batches are safe to use concurrently from multiple goroutines.

func (*Batch) Close 

func (batch *Batch) Close() error

Close closes the batch, releasing the connection lock and returning an error if reading the batch failed for any reason.

func (*Batch) HighWaterMark 

func (batch *Batch) HighWaterMark() int64

Watermark returns the current highest watermark in a partition.

func (*Batch) Offset 

func (batch *Batch) Offset() int64

Offset returns the offset of the next message in the batch.

func (*Batch) Read 

func (batch *Batch) Read(b []byte) (int, error)

Read reads the value of the next message from the batch into b, returning the number of bytes read, or an error if the next message couldn't be read.

If an error is returned the batch cannot be used anymore and calling Read again will keep returning that error. All errors except io.EOF (indicating that the program consumed all messages from the batch) are also returned by Close.

The method fails with io.ErrShortBuffer if the buffer passed as argument is too small to hold the message value.

func (*Batch) ReadMessage 

func (batch *Batch) ReadMessage() (Message, error)

ReadMessage reads and return the next message from the batch.

Because this method allocate memory buffers for the message key and value it is less memory-efficient than Read, but has the advantage of never failing with io.ErrShortBuffer.

func (*Batch) Throttle 

func (batch *Batch) Throttle() time.Duration

Throttle gives the throttling duration applied by the kafka server on the connection.

type Broker 

type Broker struct {
	Host string
	Port int
	ID   int
}

Broker carries the metadata associated with a kafka broker.

type CompressionCodec 

type CompressionCodec interface {
	// Code returns the compression codec code
	Code() int8

	// Encode encodes the src data and writes the result to dst.
	// If ths destination buffer is too small, the function should
	// return the bytes.ErrToolarge error.
	Encode(dst, src []byte) (int, error)

	// Decode decodes the src data and writes the result to dst.
	// If ths destination buffer is too small, the function should
	// return the bytes.ErrToolarge error.
	Decode(dst, src []byte) (int, error)
}

CompressionCodec represents a compression codec to encode and decode the messages. See : https://cwiki.apache.org/confluence/display/KAFKA/Compression

type ConfigEntry 

type ConfigEntry struct {
	ConfigName  string
	ConfigValue string
}

type Conn 

type Conn struct {
	// contains filtered or unexported fields
}

Conn represents a connection to a kafka broker.

Instances of Conn are safe to use concurrently from multiple goroutines.

func Dial 

func Dial(network string, address string) (*Conn, error)

Dial is a convenience wrapper for DefaultDialer.Dial.

func DialContext 

func DialContext(ctx context.Context, network string, address string) (*Conn, error)

DialContext is a convenience wrapper for DefaultDialer.DialContext.

func DialLeader 

func DialLeader(ctx context.Context, network string, address string, topic string, partition int) (*Conn, error)

DialLeader is a convenience wrapper for DefaultDialer.DialLeader.

func NewConn 

func NewConn(conn net.Conn, topic string, partition int) *Conn

NewConn returns a new kafka connection for the given topic and partition.

func NewConnWith 

func NewConnWith(conn net.Conn, config ConnConfig) *Conn

NewConnWith returns a new kafka connection configured with config.

func (*Conn) Close 

func (c *Conn) Close() error

Close closes the kafka connection.

func (*Conn) CreateTopics 

func (c *Conn) CreateTopics(topics ...TopicConfig) error

CreateTopics creates one topic per provided configuration with idempotent operational semantics. In other words, if CreateTopics is invoked with a configuration for an existing topic, it will have no effect.

func (*Conn) DeleteTopics 

func (c *Conn) DeleteTopics(topics ...string) error

DeleteTopics deletes the specified topics.

func (*Conn) LocalAddr 

func (c *Conn) LocalAddr() net.Addr

LocalAddr returns the local network address.

func (*Conn) Offset 

func (c *Conn) Offset() (offset int64, whence int)

Offset returns the current offset of the connection as pair of integers, where the first one is an offset value and the second one indicates how to interpret it.

See Seek for more details about the offset and whence values.

func (*Conn) Read 

func (c *Conn) Read(b []byte) (int, error)

Read reads the message at the current offset from the connection, advancing the offset on success so the next call to a read method will produce the next message. The method returns the number of bytes read, or an error if something went wrong.

While it is safe to call Read concurrently from multiple goroutines it may be hard for the program to predict the results as the connection offset will be read and written by multiple goroutines, they could read duplicates, or messages may be seen by only some of the goroutines.

The method fails with io.ErrShortBuffer if the buffer passed as argument is too small to hold the message value.

This method is provided to satisfy the net.Conn interface but is much less efficient than using the more general purpose ReadBatch method.

func (*Conn) ReadBatch 

func (c *Conn) ReadBatch(minBytes int, maxBytes int) *Batch

ReadBatch reads a batch of messages from the kafka server. The method always returns a non-nil Batch value. If an error occurred, either sending the fetch request or reading the response, the error will be made available by the returned value of the batch's Close method.

While it is safe to call ReadBatch concurrently from multiple goroutines it may be hard for the program to predict the results as the connection offset will be read and written by multiple goroutines, they could read duplicates, or messages may be seen by only some of the goroutines.

A program doesn't specify the number of messages in wants from a batch, but gives the minimum and maximum number of bytes that it wants to receive from the kafka server.

func (*Conn) ReadFirstOffset 

func (c *Conn) ReadFirstOffset() (int64, error)

ReadFirstOffset returns the first offset available on the connection.

func (*Conn) ReadLastOffset 

func (c *Conn) ReadLastOffset() (int64, error)

ReadLastOffset returns the last offset available on the connection.

func (*Conn) ReadMessage 

func (c *Conn) ReadMessage(maxBytes int) (Message, error)

ReadMessage reads the message at the current offset from the connection, advancing the offset on success so the next call to a read method will produce the next message.

Because this method allocate memory buffers for the message key and value it is less memory-efficient than Read, but has the advantage of never failing with io.ErrShortBuffer.

While it is safe to call Read concurrently from multiple goroutines it may be hard for the program to predict the results as the connection offset will be read and written by multiple goroutines, they could read duplicates, or messages may be seen by only some of the goroutines.

This method is provided for convenience purposes but is much less efficient than using the more general purpose ReadBatch method.

func (*Conn) ReadOffset 

func (c *Conn) ReadOffset(t time.Time) (int64, error)

ReadOffset returns the offset of the first message with a timestamp equal or greater to t.

func (*Conn) ReadOffsets 

func (c *Conn) ReadOffsets() (first int64, last int64, err error)

ReadOffsets returns the absolute first and last offsets of the topic used by the connection.

func (*Conn) ReadPartitions 

func (c *Conn) ReadPartitions(topics ...string) (partitions []Partition, err error)

ReadPartitions returns the list of available partitions for the given list of topics.

If the method is called with no topic, it uses the topic configured on the connection. If there are none, the method fetches all partitions of the kafka cluster.

func (*Conn) RemoteAddr 

func (c *Conn) RemoteAddr() net.Addr

RemoteAddr returns the remote network address.

func (*Conn) Seek 

func (c *Conn) Seek(offset int64, whence int) (int64, error)

Seek sets the offset for the next read or write operation according to whence, which should be one of SeekStart, SeekAbsolute, SeekEnd, or SeekCurrent. When seeking relative to the end, the offset is subtracted from the current offset. Note that for historical reasons, these do not align with the usual whence constants as in lseek(2) or os.Seek. The method returns the new absolute offset of the connection.

func (*Conn) SetDeadline 

func (c *Conn) SetDeadline(t time.Time) error

SetDeadline sets the read and write deadlines associated with the connection. It is equivalent to calling both SetReadDeadline and SetWriteDeadline.

A deadline is an absolute time after which I/O operations fail with a timeout (see type Error) instead of blocking. The deadline applies to all future and pending I/O, not just the immediately following call to Read or Write. After a deadline has been exceeded, the connection may be closed if it was found to be in an unrecoverable state.

A zero value for t means I/O operations will not time out.

func (*Conn) SetReadDeadline 

func (c *Conn) SetReadDeadline(t time.Time) error

SetReadDeadline sets the deadline for future Read calls and any currently-blocked Read call. A zero value for t means Read will not time out.

func (*Conn) SetRequiredAcks 

func (c *Conn) SetRequiredAcks(n int) error

SetRequiredAcks sets the number of acknowledges from replicas that the connection requests when producing messages.

func (*Conn) SetWriteDeadline 

func (c *Conn) SetWriteDeadline(t time.Time) error

SetWriteDeadline sets the deadline for future Write calls and any currently-blocked Write call. Even if write times out, it may return n > 0, indicating that some of the data was successfully written. A zero value for t means Write will not time out.

func (*Conn) Write 

func (c *Conn) Write(b []byte) (int, error)

Write writes a message to the kafka broker that this connection was established to. The method returns the number of bytes written, or an error if something went wrong.

The operation either succeeds or fail, it never partially writes the message.

This method is exposed to satisfy the net.Conn interface but is less efficient than the more general purpose WriteMessages method.

func (*Conn) WriteMessages 

func (c *Conn) WriteMessages(msgs ...Message) (int, error)

WriteMessages writes a batch of messages to the connection's topic and partition, returning the number of bytes written. The write is an atomic operation, it either fully succeeds or fails.

type ConnConfig 

type ConnConfig struct {
	ClientID  string
	Topic     string
	Partition int
}

ConnConfig is a configuration object used to create new instances of Conn.

type Dialer 

type Dialer struct {
	// Unique identifier for client connections established by this Dialer.
	ClientID string

	// Timeout is the maximum amount of time a dial will wait for a connect to
	// complete. If Deadline is also set, it may fail earlier.
	//
	// The default is no timeout.
	//
	// When dialing a name with multiple IP addresses, the timeout may be
	// divided between them.
	//
	// With or without a timeout, the operating system may impose its own
	// earlier timeout. For instance, TCP timeouts are often around 3 minutes.
	Timeout time.Duration

	// Deadline is the absolute point in time after which dials will fail.
	// If Timeout is set, it may fail earlier.
	// Zero means no deadline, or dependent on the operating system as with the
	// Timeout option.
	Deadline time.Time

	// LocalAddr is the local address to use when dialing an address.
	// The address must be of a compatible type for the network being dialed.
	// If nil, a local address is automatically chosen.
	LocalAddr net.Addr

	// DualStack enables RFC 6555-compliant "Happy Eyeballs" dialing when the
	// network is "tcp" and the destination is a host name with both IPv4 and
	// IPv6 addresses. This allows a client to tolerate networks where one
	// address family is silently broken.
	DualStack bool

	// FallbackDelay specifies the length of time to wait before spawning a
	// fallback connection, when DualStack is enabled.
	// If zero, a default delay of 300ms is used.
	FallbackDelay time.Duration

	// KeepAlive specifies the keep-alive period for an active network
	// connection.
	// If zero, keep-alives are not enabled. Network protocols that do not
	// support keep-alives ignore this field.
	KeepAlive time.Duration

	// Resolver optionally specifies an alternate resolver to use.
	Resolver Resolver

	// TLS enables Dialer to open secure connections.  If nil, standard net.Conn
	// will be used.
	TLS *tls.Config
}

The Dialer type mirrors the net.Dialer API but is designed to open kafka connections instead of raw network connections.

func (*Dialer) Dial 

func (d *Dialer) Dial(network string, address string) (*Conn, error)

Dial connects to the address on the named network.

func (*Dialer) DialContext 

func (d *Dialer) DialContext(ctx context.Context, network string, address string) (*Conn, error)

DialContext connects to the address on the named network using the provided context.

The provided Context must be non-nil. If the context expires before the connection is complete, an error is returned. Once successfully connected, any expiration of the context will not affect the connection.

When using TCP, and the host in the address parameter resolves to multiple network addresses, any dial timeout (from d.Timeout or ctx) is spread over each consecutive dial, such that each is given an appropriate fraction of the time to connect. For example, if a host has 4 IP addresses and the timeout is 1 minute, the connect to each single address will be given 15 seconds to complete before trying the next one.

func (*Dialer) DialLeader 

func (d *Dialer) DialLeader(ctx context.Context, network string, address string, topic string, partition int) (*Conn, error)

DialLeader opens a connection to the leader of the partition for a given topic.

The address given to the DialContext method may not be the one that the connection will end up being established to, because the dialer will lookup the partition leader for the topic and return a connection to that server. The original address is only used as a mechanism to discover the configuration of the kafka cluster that we're connecting to.

func (*Dialer) LookupLeader 

func (d *Dialer) LookupLeader(ctx context.Context, network string, address string, topic string, partition int) (Broker, error)

LookupLeader searches for the kafka broker that is the leader of the partition for a given topic, returning a Broker value representing it.

func (*Dialer) LookupPartitions 

func (d *Dialer) LookupPartitions(ctx context.Context, network string, address string, topic string) ([]Partition, error)

LookupPartitions returns the list of partitions that exist for the given topic.

type DurationStats 

type DurationStats struct {
	Avg time.Duration `metric:"avg" type:"gauge"`
	Min time.Duration `metric:"min" type:"gauge"`
	Max time.Duration `metric:"max" type:"gauge"`
}

DurationStats is a data structure that carries a summary of observed duration values. The average, minimum, and maximum are reported.

type Error 

type Error int

Error represents the different error codes that may be returned by kafka. 

const (
	Unknown                            Error = -1
	OffsetOutOfRange                   Error = 1
	InvalidMessage                     Error = 2
	UnknownTopicOrPartition            Error = 3
	InvalidMessageSize                 Error = 4
	LeaderNotAvailable                 Error = 5
	NotLeaderForPartition              Error = 6
	RequestTimedOut                    Error = 7
	BrokerNotAvailable                 Error = 8
	ReplicaNotAvailable                Error = 9
	MessageSizeTooLarge                Error = 10
	StaleControllerEpoch               Error = 11
	OffsetMetadataTooLarge             Error = 12
	GroupLoadInProgress                Error = 14
	GroupCoordinatorNotAvailable       Error = 15
	NotCoordinatorForGroup             Error = 16
	InvalidTopic                       Error = 17
	RecordListTooLarge                 Error = 18
	NotEnoughReplicas                  Error = 19
	NotEnoughReplicasAfterAppend       Error = 20
	InvalidRequiredAcks                Error = 21
	IllegalGeneration                  Error = 22
	InconsistentGroupProtocol          Error = 23
	InvalidGroupId                     Error = 24
	UnknownMemberId                    Error = 25
	InvalidSessionTimeout              Error = 26
	RebalanceInProgress                Error = 27
	InvalidCommitOffsetSize            Error = 28
	TopicAuthorizationFailed           Error = 29
	GroupAuthorizationFailed           Error = 30
	ClusterAuthorizationFailed         Error = 31
	InvalidTimestamp                   Error = 32
	UnsupportedSASLMechanism           Error = 33
	IllegalSASLState                   Error = 34
	UnsupportedVersion                 Error = 35
	TopicAlreadyExists                 Error = 36
	InvalidPartitionNumber             Error = 37
	InvalidReplicationFactor           Error = 38
	InvalidReplicaAssignment           Error = 39
	InvalidConfiguration               Error = 40
	NotController                      Error = 41
	InvalidRequest                     Error = 42
	UnsupportedForMessageFormat        Error = 43
	PolicyViolation                    Error = 44
	OutOfOrderSequenceNumber           Error = 45
	DuplicateSequenceNumber            Error = 46
	InvalidProducerEpoch               Error = 47
	InvalidTransactionState            Error = 48
	InvalidProducerIDMapping           Error = 49
	InvalidTransactionTimeout          Error = 50
	ConcurrentTransactions             Error = 51
	TransactionCoordinatorFenced       Error = 52
	TransactionalIDAuthorizationFailed Error = 53
	SecurityDisabled                   Error = 54
	BrokerAuthorizationFailed          Error = 55
)

func (Error) Description 

func (e Error) Description() string

Description returns a human readable description of cause of the error.

func (Error) Error 

func (e Error) Error() string

Error satisfies the error interface.

func (Error) Temporary 

func (e Error) Temporary() bool

Temporary returns true if the operation that generated the error may succeed if retried at a later time.

func (Error) Timeout 

func (e Error) Timeout() bool

Timeout returns true if the error was due to a timeout.

func (Error) Title 

func (e Error) Title() string

Title returns a human readable title for the error.

type Hash 

type Hash struct {
	Hasher hash.Hash32
	// contains filtered or unexported fields
}

Hash is a Balancer that uses the provided hash function to determine which partition to route messages to. This ensures that messages with the same key are routed to the same partition.

The logic to calculate the partition is: 

hasher.Sum32() % len(partitions) => partition

By default, Hash uses the FNV-1a algorithm. This is the same algorithm used by the Sarama Producer and ensures that messages produced by kafka-go will be delivered to the same topics that the Sarama producer would be delivered to

func (*Hash) Balance 

func (h *Hash) Balance(msg Message, partitions ...int) (partition int)

type LeastBytes 

type LeastBytes struct {
	// contains filtered or unexported fields
}

LeastBytes is a Balancer implementation that routes messages to the partition that has received the least amount of data.

Note that no coordination is done between multiple producers, having good balancing relies on the fact that each producer using a LeastBytes balancer should produce well balanced messages.

func (*LeastBytes) Balance 

func (lb *LeastBytes) Balance(msg Message, partitions ...int) int

Balance satisfies the Balancer interface.

type ListGroupsResponseGroupV1 

type ListGroupsResponseGroupV1 struct {
	// GroupID holds the unique group identifier
	GroupID      string
	ProtocolType string
}

type Message 

type Message struct {
	// Topic is reads only and MUST NOT be set when writing messages
	Topic string

	// Partition is reads only and MUST NOT be set when writing messages
	Partition int
	Offset    int64
	Key       []byte
	Value     []byte

	// If not set at the creation, Time will be automatically set when
	// writing the message.
	Time time.Time

	// Compression codec used to encode the message value
	CompressionCodec
}

Message is a data structure representing kafka messages.

type Partition 

type Partition struct {
	Topic    string
	Leader   Broker
	Replicas []Broker
	Isr      []Broker
	ID       int
}

Partition carries the metadata associated with a kafka partition.

type Reader 

type Reader struct {
	// contains filtered or unexported fields
}

Reader provides a high-level API for consuming messages from kafka.

A Reader automatically manages reconnections to a kafka server, and blocking methods have context support for asynchronous cancellations.

func NewReader 

func NewReader(config ReaderConfig) *Reader

NewReader creates and returns a new Reader configured with config.

func (*Reader) Close 

func (r *Reader) Close() error

Close closes the stream, preventing the program from reading any more messages from it.

func (*Reader) CommitMessages 

func (r *Reader) CommitMessages(ctx context.Context, msgs ...Message) error

CommitMessages commits the list of messages passed as argument. The program may pass a context to asynchronously cancel the commit operation when it was configured to be blocking.

func (*Reader) Config 

func (r *Reader) Config() ReaderConfig

Config returns the reader's configuration.

func (*Reader) FetchMessage 

func (r *Reader) FetchMessage(ctx context.Context) (Message, error)

FetchMessage reads and return the next message from the r. The method call blocks until a message becomes available, or an error occurs. The program may also specify a context to asynchronously cancel the blocking operation.

The method returns io.EOF to indicate that the reader has been closed.

FetchMessage does not commit offsets automatically when using consumer groups. Use CommitMessages to commit the offset.

func (*Reader) Lag 

func (r *Reader) Lag() int64

Lag returns the lag of the last message returned by ReadMessage.

func (*Reader) Offset 

func (r *Reader) Offset() int64

Offset returns the current offset of the reader.

func (*Reader) ReadLag 

func (r *Reader) ReadLag(ctx context.Context) (lag int64, err error)

ReadLag returns the current lag of the reader by fetching the last offset of the topic and partition and computing the difference between that value and the offset of the last message returned by ReadMessage.

This method is intended to be used in cases where a program may be unable to call ReadMessage to update the value returned by Lag, but still needs to get an up to date estimation of how far behind the reader is. For example when the consumer is not ready to process the next message.

The function returns a lag of zero when the reader's current offset is negative.

func (*Reader) ReadMessage 

func (r *Reader) ReadMessage(ctx context.Context) (Message, error)

ReadMessage reads and return the next message from the r. The method call blocks until a message becomes available, or an error occurs. The program may also specify a context to asynchronously cancel the blocking operation.

The method returns io.EOF to indicate that the reader has been closed.

If consumer groups are used, ReadMessage will automatically commit the offset when called.

func (*Reader) SetOffset 

func (r *Reader) SetOffset(offset int64) error

SetOffset changes the offset from which the next batch of messages will be read.

Setting the offset ot -1 means to seek to the first offset. Setting the offset to -2 means to seek to the last offset.

The method fails with io.ErrClosedPipe if the reader has already been closed.

func (*Reader) Stats 

func (r *Reader) Stats() ReaderStats

Stats returns a snapshot of the reader stats since the last time the method was called, or since the reader was created if it is called for the first time.

A typical use of this method is to spawn a goroutine that will periodically call Stats on a kafka reader and report the metrics to a stats collection system.

type ReaderConfig 

type ReaderConfig struct {
	// The list of broker addresses used to connect to the kafka cluster.
	Brokers []string

	// GroupID holds the optional consumer group id.  If GroupID is specified, then
	// Partition should NOT be specified e.g. 0
	GroupID string

	// The topic to read messages from.
	Topic string

	// Partition to read messages from.  Either Partition or GroupID may
	// be assigned, but not both
	Partition int

	// An dialer used to open connections to the kafka server. This field is
	// optional, if nil, the default dialer is used instead.
	Dialer *Dialer

	// The capacity of the internal message queue, defaults to 100 if none is
	// set.
	QueueCapacity int

	// Min and max number of bytes to fetch from kafka in each request.
	MinBytes int
	MaxBytes int

	// Maximum amount of time to wait for new data to come when fetching batches
	// of messages from kafka.
	MaxWait time.Duration

	// ReadLagInterval sets the frequency at which the reader lag is updated.
	// Setting this field to a negative value disables lag reporting.
	ReadLagInterval time.Duration

	// HeartbeatInterval sets the optional frequency at which the reader sends the consumer
	// group heartbeat update.
	//
	// Default: 3s
	//
	// Only used when GroupID is set
	HeartbeatInterval time.Duration

	// CommitInterval indicates the interval at which offsets are committed to
	// the broker.  If 0, commits will be handled synchronously.
	//
	// Defaults to 1s
	//
	// Only used when GroupID is set
	CommitInterval time.Duration

	// SessionTimeout optionally sets the length of time that may pass without a heartbeat
	// before the coordinator considers the consumer dead and initiates a rebalance.
	//
	// Default: 30s
	//
	// Only used when GroupID is set
	SessionTimeout time.Duration

	// RebalanceTimeout optionally sets the length of time the coordinator will wait
	// for members to join as part of a rebalance.  For kafka servers under higher
	// load, it may be useful to set this value higher.
	//
	// Default: 30s
	//
	// Only used when GroupID is set
	RebalanceTimeout time.Duration

	// RetentionTime optionally sets the length of time the consumer group will be saved
	// by the broker
	//
	// Default: 24h
	//
	// Only used when GroupID is set
	RetentionTime time.Duration

	// If not nil, specifies a logger used to report internal changes within the
	// reader.
	Logger *log.Logger

	// ErrorLogger is the logger used to report errors. If nil, the reader falls
	// back to using Logger instead.
	ErrorLogger *log.Logger
}

ReaderConfig is a configuration object used to create new instances of Reader.

type ReaderStats 

type ReaderStats struct {
	Dials      int64 `metric:"kafka.reader.dial.count"      type:"counter"`
	Fetches    int64 `metric:"kafak.reader.fetch.count"     type:"counter"` // typo here, but I'm reluctant to fix it
	Messages   int64 `metric:"kafka.reader.message.count"   type:"counter"`
	Bytes      int64 `metric:"kafka.reader.message.bytes"   type:"counter"`
	Rebalances int64 `metric:"kafka.reader.rebalance.count" type:"counter"`
	Timeouts   int64 `metric:"kafka.reader.timeout.count"   type:"counter"`
	Errors     int64 `metric:"kafka.reader.error.count"     type:"counter"`

	DialTime   DurationStats `metric:"kafka.reader.dial.seconds"`
	ReadTime   DurationStats `metric:"kafka.reader.read.seconds"`
	WaitTime   DurationStats `metric:"kafka.reader.wait.seconds"`
	FetchSize  SummaryStats  `metric:"kafka.reader.fetch.size"`
	FetchBytes SummaryStats  `metric:"kafka.reader.fetch.bytes"`

	Offset        int64         `metric:"kafka.reader.offset"          type:"gauge"`
	Lag           int64         `metric:"kafka.reader.lag"             type:"gauge"`
	MinBytes      int64         `metric:"kafka.reader.fetch_bytes.min" type:"gauge"`
	MaxBytes      int64         `metric:"kafka.reader.fetch_bytes.max" type:"gauge"`
	MaxWait       time.Duration `metric:"kafka.reader.fetch_wait.max"  type:"gauge"`
	QueueLength   int64         `metric:"kafka.reader.queue.length"    type:"gauge"`
	QueueCapacity int64         `metric:"kafka.reader.queue.capacity"  type:"gauge"`

	ClientID  string `tag:"client_id"`
	Topic     string `tag:"topic"`
	Partition string `tag:"partition"`
}

ReaderStats is a data structure returned by a call to Reader.Stats that exposes details about the behavior of the reader.

type ReplicaAssignment 

type ReplicaAssignment struct {
	Partition int
	Replicas  int
}

type Resolver 

type Resolver interface {
	// LookupHost looks up the given host using the local resolver.
	// It returns a slice of that host's addresses.
	LookupHost(ctx context.Context, host string) (addrs []string, err error)
}

The Resolver interface is used as an abstraction to provide service discovery of the hosts of a kafka cluster.

type RoundRobin 

type RoundRobin struct {
	// contains filtered or unexported fields
}

RoundRobin is an Balancer implementation that equally distributes messages across all available partitions.

func (*RoundRobin) Balance 

func (rr *RoundRobin) Balance(msg Message, partitions ...int) int

Balance satisfies the Balancer interface.

type SummaryStats 

type SummaryStats struct {
	Avg int64 `metric:"avg" type:"gauge"`
	Min int64 `metric:"min" type:"gauge"`
	Max int64 `metric:"max" type:"gauge"`
}

SummaryStats is a data structure that carries a summary of observed values. The average, minimum, and maximum are reported.

type TopicConfig 

type TopicConfig struct {
	// Topic name
	Topic string

	// NumPartitions created. -1 indicates unset.
	NumPartitions int

	// ReplicationFactor for the topic. -1 indicates unset.
	ReplicationFactor int

	// ReplicaAssignments among kafka brokers for this topic partitions. If this
	// is set num_partitions and replication_factor must be unset.
	ReplicaAssignments []ReplicaAssignment

	// ConfigEntries holds topic level configuration for topic to be set.
	ConfigEntries []ConfigEntry
}

type Writer 

type Writer struct {
	// contains filtered or unexported fields
}

The Writer type provides the implementation of a producer of kafka messages that automatically distributes messages across partitions of a single topic using a configurable balancing policy.

Instances of Writer are safe to use concurrently from multiple goroutines.

Example 
w := kafka.NewWriter(kafka.WriterConfig{
	Brokers: []string{"localhost:9092"},
	Topic:   "Topic-1",
})

w.WriteMessages(context.Background(),
	kafka.Message{
		Key:   []byte("Key-A"),
		Value: []byte("Hello World!"),
	},
)

w.Close()

Output:

func NewWriter 

func NewWriter(config WriterConfig) *Writer

NewWriter creates and returns a new Writer configured with config.

func (*Writer) Close 

func (w *Writer) Close() (err error)

Close flushes all buffered messages and closes the writer. The call to Close aborts any concurrent calls to WriteMessages, which then return with the io.ErrClosedPipe error.

func (*Writer) Stats 

func (w *Writer) Stats() WriterStats

Stats returns a snapshot of the writer stats since the last time the method was called, or since the writer was created if it is called for the first time.

A typical use of this method is to spawn a goroutine that will periodically call Stats on a kafka writer and report the metrics to a stats collection system.

func (*Writer) WriteMessages 

func (w *Writer) WriteMessages(ctx context.Context, msgs ...Message) error

WriteMessages writes a batch of messages to the kafka topic configured on this writer.

Unless the writer was configured to write messages asynchronously, the method blocks until all messages have been written, or until the maximum number of attempts was reached.

When the method returns an error, there's no way to know yet which messages have succeeded of failed.

The context passed as first argument may also be used to asynchronously cancel the operation. Note that in this case there are no guarantees made on whether messages were written to kafka. The program should assume that the whole batch failed and re-write the messages later (which could then cause duplicates).

type WriterConfig 

type WriterConfig struct {
	// The list of brokers used to discover the partitions available on the
	// kafka cluster.
	//
	// This field is required, attempting to create a writer with an empty list
	// of brokers will panic.
	Brokers []string

	// The topic that the writer will produce messages to.
	//
	// This field is required, attempting to create a writer with an empty topic
	// will panic.
	Topic string

	// The dialer used by the writer to establish connections to the kafka
	// cluster.
	//
	// If nil, the default dialer is used instead.
	Dialer *Dialer

	// The balancer used to distribute messages across partitions.
	//
	// The default is to use a round-robin distribution.
	Balancer Balancer

	// Limit on how many attempts will be made to deliver a message.
	//
	// The default is to try at most 10 times.
	MaxAttempts int

	// A hint on the capacity of the writer's internal message queue.
	//
	// The default is to use a queue capacity of 100 messages.
	QueueCapacity int

	// Limit on how many messages will be buffered before being sent to a
	// partition.
	//
	// The default is to use a target batch size of 100 messages.
	BatchSize int

	// Time limit on how often incomplete message batches will be flushed to
	// kafka.
	//
	// The default is to flush at least every second.
	BatchTimeout time.Duration

	// Timeout for read operations performed by the Writer.
	//
	// Defaults to 10 seconds.
	ReadTimeout time.Duration

	// Timeout for write operation performed by the Writer.
	//
	// Defaults to 10 seconds.
	WriteTimeout time.Duration

	// This interval defines how often the list of partitions is refreshed from
	// kafka. It allows the writer to automatically handle when new partitions
	// are added to a topic.
	//
	// The default is to refresh partitions every 15 seconds.
	RebalanceInterval time.Duration

	// Number of acknowledges from partition replicas required before receiving
	// a response to a produce request (default to -1, which means to wait for
	// all replicas).
	RequiredAcks int

	// Setting this flag to true causes the WriteMessages method to never block.
	// It also means that errors are ignored since the caller will not receive
	// the returned value. Use this only if you don't care about guarantees of
	// whether the messages were written to kafka.
	Async bool

	// CompressionCodec set the codec to be used to compress Kafka messages.
	// Note that messages are allowed to overwrite the compression codec individualy.
	CompressionCodec
	// contains filtered or unexported fields
}

WriterConfig is a configuration type used to create new instances of Writer.

type WriterStats 

type WriterStats struct {
	Dials      int64 `metric:"kafka.writer.dial.count"      type:"counter"`
	Writes     int64 `metric:"kafka.writer.write.count"     type:"counter"`
	Messages   int64 `metric:"kafka.writer.message.count"   type:"counter"`
	Bytes      int64 `metric:"kafka.writer.message.bytes"   type:"counter"`
	Rebalances int64 `metric:"kafka.writer.rebalance.count" type:"counter"`
	Errors     int64 `metric:"kafka.writer.error.count"     type:"counter"`

	DialTime  DurationStats `metric:"kafka.writer.dial.seconds"`
	WriteTime DurationStats `metric:"kafka.writer.write.seconds"`
	WaitTime  DurationStats `metric:"kafka.writer.wait.seconds"`
	Retries   SummaryStats  `metric:"kafka.writer.retries.count"`
	BatchSize SummaryStats  `metric:"kafka.writer.batch.size"`

	MaxAttempts       int64         `metric:"kafka.writer.attempts.max"       type:"gauge"`
	MaxBatchSize      int64         `metric:"kafka.writer.batch.max"          type:"gauge"`
	BatchTimeout      time.Duration `metric:"kafka.writer.batch.timeout"      type:"gauge"`
	ReadTimeout       time.Duration `metric:"kafka.writer.read.timeout"       type:"gauge"`
	WriteTimeout      time.Duration `metric:"kafka.writer.write.timeout"      type:"gauge"`
	RebalanceInterval time.Duration `metric:"kafka.writer.rebalance.interval" type:"gauge"`
	RequiredAcks      int64         `metric:"kafka.writer.acks.required"      type:"gauge"`
	Async             bool          `metric:"kafka.writer.async"              type:"gauge"`
	QueueLength       int64         `metric:"kafka.writer.queue.length"       type:"gauge"`
	QueueCapacity     int64         `metric:"kafka.writer.queue.capacity"     type:"gauge"`

	ClientID string `tag:"client_id"`
	Topic    string `tag:"topic"`
}

WriterStats is a data structure returned by a call to Writer.Stats that exposes details about the behavior of the writer.

Source Files

View all Source files

Directories

Path Synopsis

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.