- Building and Installing
- Linking with an Application
- API v1 Summary dafka_consumer - no title found dafka_producer - no title found dafka_store - no title found dafka_tower - no title found
- Documentation
- Development
- Hints to Contributors
- Code Generation
- This Document
Dafka is a decentralize distributed streaming platform. What exactly does that mean?
A streaming platform has three key capabilities:
- Publish and subscribe to streams of records, similar to a message queue or enterprise messaging system.
- Store streams of records in a fault-tolerant durable way.
- Process streams of records as they occur.
Dafka is generally used for two broad classes of applications:
- Building real-time streaming data pipelines that reliably get data between systems or applications
- Building real-time streaming applications that transform or react to the streams of data
To understand how Dafka does these things, let's dive in and explore Dafka's capabilities from the bottom up.
First a few concepts:
- Dafka is run as a cluster on one or more servers.
- The Dafka cluster stores streams of records in categories called topics.
- Each record consists of a arbitrary value.
- Producers send record to the Cluster and directly to the Consumers.
- Missed records are obtained either from the Producer or the Cluster.
In Dafka the communication between clients is done with a simple, high-performance, language and transport agnostic protocol. This protocol is versioned and maintains backwards compatibility with older version. We provide a C and Java client for Dafka.
Dafka provides an abstraction for records called topic.
A topic is a name to which records are published. Topics in Dafka are always multi-subscriber; that is, a topic can have zero, one, or many consumers that subscribe to the records written to it.
Each Dafka topic consists of at least one partitions that looks like this:
Each partition is an ordered, immutable sequence of records that is continually appended to. The records in the partitions are each assigned a sequential id number called the offset that uniquely identifies each record within the partition.
The Dafka cluster durably persists all published records — whether or not they have been consumed.
Consumers maintain their own offset while reading records of a partition. In fact neither the Dafka Cluster nor the producers keep track of the consumers offset. This design allows Consumer to either reset their offset to an older offset and re-read records or set their offset to a newer offset and skip ahead.
In that way consumer have no influence on the cluster, the producer and other consumers. They simply can come and go as they please.
Partitions are distributed to the Dafka Cluster which consists of Dafka Stores. Each partition is replicated to each store for fault tolerance.
Producers publish records to a topic. Each producer creates its own partition that only it publishes to. Records are send directly to stores and consumers. When a producer goes offline its partition is still available to consumers from the Dafka stores.
Consumers subscribe to a topic. Each consumer will receive records published to that topic from all partitions.
Each Dafka cluster has one or more towers. The towers are used to connect producer, consumers and stores to each other. At this point no traffic is proxied through the towers.
Dafka gives the following guarantees:
- Records sent by a producer are appended in the stores in the same order they are sent.
- Consumers will provide records of a partition to the user in the same order they are sent by the producer.
We designed Dafka to be a drop-in replacement for Apache Kafka.
While Kafka makes it easy for consumers to come and go as they like, their consumer group feature which relies on finding consensus in a group of peers makes joining very expensive. It can take seconds before a consumer is ready to consume records. The same is true for producer. Dafka tries to avoid finding consensus and therefore intentionally avoids features like consumer groups in favor of higher throughput, lower latency as well as faster consumer and producer initialization.
This design section discusses the different message types of the Dafka protocol.
Producers published records using the RECORD message type. RECORD messages are send directly to all connected stores as well as all connected consumers. Once a producer published its first records it starts sending HEAD messages at a regular interval informing both stores and consumer about the last published records which gives stores and consumers a chance to figure out whether or not the missed one or more records.
Because producers publish records directly to consumers the presence of a store is not necessarily required. When a new consumer joins they can request the producers to supply all already published records. Therefore the producer must store all published records that are not stored by a configurable minimum number stores. To inform a producer about the successful storing of a records the stores send a ACK message to the producer.
Consumer will only start listening for HEAD message once they subscribed for a topic. Whenever a new subscription created by a consumer it is not enough to listen to producers HEAD messages to catch up upon the current offset of their partition. For one there's a time penalty until producers HEAD intervals triggered and more severe a producer may already have disappeared. Hence consumers will send a GET-HEADS message to the stores to request the offset for each partition they stored for a topic.
As a response each stores will answer with DIRECT-HEAD messages each containing the offset for a partition.
Consumer can discover missed records by either receiving HEAD messages or receiving a RECORD messages with a higher offset than they currently have for a certain partition. In order to fetch missed messages consumers send a FETCH message to all connected stores and the producer of that message to request the missed messages.
As a response to a FETCH message a store and/or producer may send all missed records that the consumer requested directly to the consumer with the DIRECT-RECORD.
The implementation is documented in RFC 46/DAFKA
The contributors are listed in AUTHORS. This project uses the MPL v2 license, see LICENSE.
Dafka uses the C4.1 (Collective Code Construction Contract) process for contributions.
Dafka uses the CLASS (C Language Style for Scalability) guide for code style.
To report an issue, use the Dafka issue tracker at github.com.
To start with, you need at least these packages:
-
{{git-all}} -- git is how we share code with other people.
-
{{build-essential}}, {{libtool}}, {{pkg-config}} - the C compiler and related tools.
-
{{autotools-dev}}, {{autoconf}}, {{automake}} - the GNU autoconf makefile generators.
-
{{cmake}} - the CMake makefile generators (an alternative to autoconf).
Plus some others:
-
{{uuid-dev}}, {{libpcre3-dev}} - utility libraries.
-
{{valgrind}} - a useful tool for checking your code.
-
{{pkg-config}} - an optional useful tool to make building with dependencies easier.
Which we install like this (using the Debian-style apt-get package manager):
sudo apt-get update
sudo apt-get install -y \
git-all build-essential libtool \
pkg-config autotools-dev autoconf automake cmake \
uuid-dev libpcre3-dev valgrind
# only execute this next line if interested in updating the man pages as well (adds to build time):
sudo apt-get install -y asciidoc
TODO...
Include dafka.h in your application and link with libdafka. Here is a typical gcc link command:
gcc myapp.c -o myapp -ldafka -lczmq -lzmq
This is the API provided by Dafka v1.x, in alphabetical order.
dafka_consumer -
TODO:
- Send earliest message when a store connects
- We must not send FETCH on every message, the problem is, that if you missed something, and there is high rate, you will end up sending a lot of fetch messages for same address
- Prioritize DIRECT_MSG messages over MSG this will avoid discrding MSGs when catching up
This is the class interface:
// This is a stable class, and may not change except for emergencies. It
// is provided in stable builds.
//
DAFKA_EXPORT void
dafka_consumer (zsock_t *pipe, void *args);
//
DAFKA_EXPORT int
dafka_consumer_subscribe (zactor_t *self, const char *subject);
// Self test of this class.
DAFKA_EXPORT void
dafka_consumer_test (bool verbose);
Please add '@interface' section in './../src/dafka_consumer.c'.
This is the class self test code:
zconfig_t *config = zconfig_new ("root", NULL);
zconfig_put (config, "test/verbose", verbose ? "1" : "0");
zconfig_put (config, "beacon/interval", "50");
zconfig_put (config, "beacon/verbose", verbose ? "1" : "0");
zconfig_put (config, "beacon/sub_address", "inproc://consumer-tower-sub");
zconfig_put (config, "beacon/pub_address", "inproc://consumer-tower-pub");
zconfig_put (config, "tower/verbose", verbose ? "1" : "0");
zconfig_put (config, "tower/sub_address", "inproc://consumer-tower-sub");
zconfig_put (config, "tower/pub_address", "inproc://consumer-tower-pub");
zconfig_put (config, "consumer/verbose", verbose ? "1" : "0");
zconfig_put (config, "producer/verbose", verbose ? "1" : "0");
zconfig_put (config, "store/verbose", verbose ? "1" : "0");
zconfig_put (config, "store/db", SELFTEST_DIR_RW "/storedb");
zactor_t *tower = zactor_new (dafka_tower_actor, config);
// -------------------------------------------------------------------
// Test with 'consumer.offset.reset = earliest' triggered by HEAD msg
// -------------------------------------------------------------------
zconfig_put (config, "consumer/offset/reset", "earliest");
zactor_t *test_peer = zactor_new (dafka_test_peer, config);
assert (test_peer);
// GIVEN a dafka consumer
zactor_t *consumer = zactor_new (dafka_consumer, config);
assert (consumer);
zclock_sleep (250); // Make sure both peers are connected to each other
// WHEN consumer subscribes to topic 'hello'
int rc = dafka_consumer_subscribe (consumer, "hello");
assert (rc == 0);
// THEN the consumer will send a GET_HEADS msg for the topic 'hello'
dafka_proto_t *msg = dafka_test_peer_recv (test_peer);
assert_get_heads_msg (msg, "hello");
// WHEN a HEAD msg with sequence larger 0 is sent on topic 'hello'
dafka_test_peer_send_head (test_peer, "hello", 1);
// THEN the consumer will send a FETCH msg for the topic 'hello'
msg = dafka_test_peer_recv (test_peer);
assert_fetch_msg (msg, "hello", 0);
// WHEN a RECORD msg with sequence 0 and content 'CONTENT' is send on topic
// 'hello'
dafka_test_peer_send_record (test_peer, "hello", 0, "CONTENT");
// THEN a consumer msg is sent to the user with topic 'hello' and content
// 'CONTENT'
dafka_consumer_msg_t *c_msg = dafka_consumer_msg_new ();
dafka_consumer_msg_recv (c_msg, consumer);
assert_consumer_msg (c_msg, "hello", "CONTENT");
dafka_consumer_msg_destroy (&c_msg);
zactor_destroy (&consumer);
zactor_destroy (&test_peer);
// ---------------------------------------------------------------------
// Test with 'consumer.offset.reset = earliest' triggered by RECORD msg
// ---------------------------------------------------------------------
zconfig_put (config, "consumer/offset/reset", "earliest");
test_peer = zactor_new (dafka_test_peer, config);
assert (test_peer);
// GIVEN a dafka consumer
consumer = zactor_new (dafka_consumer, config);
assert (consumer);
zclock_sleep (250); // Make sure both peers are connected to each other
// WHEN consumer subscribes to topic 'hello'
rc = dafka_consumer_subscribe (consumer, "hello");
assert (rc == 0);
// THEN the consumer will send a GET_HEADS msg for the topic 'hello'
msg = dafka_test_peer_recv (test_peer);
assert_get_heads_msg (msg, "hello");
// WHEN a RECORD msg with sequence larger 0 is sent on topic 'hello'
dafka_test_peer_send_record (test_peer, "hello", 1, "CONTENT");
// THEN the consumer will send a FETCH msg for the topic 'hello'
msg = dafka_test_peer_recv (test_peer);
assert_fetch_msg (msg, "hello", 0);
// WHEN a RECORD msg with sequence 0 and content 'CONTENT' is send on topic
// 'hello'
dafka_test_peer_send_record (test_peer, "hello", 0, "CONTENT");
// THEN a consumer msg is sent to the user with topic 'hello' and content
// 'CONTENT'
c_msg = dafka_consumer_msg_new ();
dafka_consumer_msg_recv (c_msg, consumer);
assert_consumer_msg (c_msg, "hello", "CONTENT");
dafka_consumer_msg_destroy (&c_msg);
zactor_destroy (&consumer);
zactor_destroy (&test_peer);
// ------------------------------------------------------------------
// Test with Producer + Store and 'consumer.offset.reset = earliest'
// ------------------------------------------------------------------
zconfig_put (config, "consumer/offset/reset", "earliest");
dafka_producer_args_t pub_args = {"hello", config};
zactor_t *producer = zactor_new (dafka_producer, &pub_args);
assert (producer);
zactor_t *store = zactor_new (dafka_store_actor, config);
assert (store);
consumer = zactor_new (dafka_consumer, config);
assert (consumer);
zclock_sleep (250);
dafka_producer_msg_t *p_msg = dafka_producer_msg_new ();
dafka_producer_msg_set_content_str (p_msg, "HELLO MATE");
rc = dafka_producer_msg_send (p_msg, producer);
assert (rc == 0);
zclock_sleep (100); // Make sure message is published before consumer subscribes
rc = dafka_consumer_subscribe (consumer, "hello");
assert (rc == 0);
zclock_sleep (250); // Make sure subscription is active before sending the next message
// This message is discarded but triggers a FETCH from the store
dafka_producer_msg_set_content_str (p_msg, "HELLO ATEM");
rc = dafka_producer_msg_send (p_msg, producer);
assert (rc == 0);
zclock_sleep (
100); // Make sure the first two messages have been received from the store and the consumer is now up to date
dafka_producer_msg_set_content_str (p_msg, "HELLO TEMA");
rc = dafka_producer_msg_send (p_msg, producer);
assert (rc == 0);
// Receive the first message from the STORE
c_msg = dafka_consumer_msg_new ();
dafka_consumer_msg_recv (c_msg, consumer);
assert_consumer_msg (c_msg, "hello", "HELLO MATE");
// Receive the second message from the STORE as the original has been discarded
dafka_consumer_msg_recv (c_msg, consumer);
assert_consumer_msg (c_msg, "hello", "HELLO ATEM");
// Receive the third message from the PUBLISHER
dafka_consumer_msg_recv (c_msg, consumer);
assert_consumer_msg (c_msg, "hello", "HELLO TEMA");
dafka_producer_msg_destroy (&p_msg);
dafka_consumer_msg_destroy (&c_msg);
zactor_destroy (&producer);
zactor_destroy (&store);
zactor_destroy (&consumer);
// --------------------------------------------------------------
// Test with Producer + Store and consumer.offset.reset = latest
// --------------------------------------------------------------
zconfig_put (config, "consumer/offset/reset", "latest");
producer = zactor_new (dafka_producer, &pub_args);
assert (producer);
consumer = zactor_new (dafka_consumer, config);
assert (consumer);
zclock_sleep (250);
// This message is missed by the consumer and later ignored because the
// offset reset is set to latest.
p_msg = dafka_producer_msg_new ();
dafka_producer_msg_set_content_str (p_msg, "HELLO MATE");
rc = dafka_producer_msg_send (p_msg, producer);
assert (rc == 0);
zclock_sleep (100); // Make sure message is published before consumer subscribes
rc = dafka_consumer_subscribe (consumer, "hello");
assert (rc == 0);
zclock_sleep (250); // Make sure subscription is active before sending the next message
dafka_producer_msg_set_content_str (p_msg, "HELLO ATEM");
rc = dafka_producer_msg_send (p_msg, producer);
assert (rc == 0);
// Receive the second message from the PRODUCER
c_msg = dafka_consumer_msg_new ();
dafka_consumer_msg_recv (c_msg, consumer);
assert_consumer_msg (c_msg, "hello", "HELLO ATEM");
// We have to create a store in-order to ack all publisher messages and allow the publisher to terminate
store = zactor_new (dafka_store_actor, config);
assert (store);
dafka_producer_msg_destroy (&p_msg);
dafka_consumer_msg_destroy (&c_msg);
zactor_destroy (&tower);
zactor_destroy (&producer);
zactor_destroy (&store);
zactor_destroy (&consumer);
zconfig_destroy (&config);dafka_publisher -
Please add '@discuss' section in './../src/dafka_producer.c'.
This is the class interface:
// This is a stable class, and may not change except for emergencies. It
// is provided in stable builds.
//
DAFKA_EXPORT void
dafka_producer (zsock_t *pipe, void *args);
//
DAFKA_EXPORT const char *
dafka_producer_address (zactor_t *self);
// Self test of this class.
DAFKA_EXPORT void
dafka_producer_test (bool verbose);
Please add '@interface' section in './../src/dafka_producer.c'.
This is the class self test code:
// Simple create/destroy test
zconfig_t *config = zconfig_new ("root", NULL);
zconfig_put (config, "beacon/verbose", verbose ? "1" : "0");
zconfig_put (config, "beacon/sub_address","inproc://producer-tower-sub");
zconfig_put (config, "beacon/pub_address","inproc://producer-tower-pub");
zconfig_put (config, "tower/verbose", verbose ? "1" : "0");
zconfig_put (config, "tower/sub_address","inproc://producer-tower-sub");
zconfig_put (config, "tower/pub_address","inproc://producer-tower-pub");
zconfig_put (config, "producer/verbose", verbose ? "1" : "0");
zconfig_put (config, "store/verbose", verbose ? "1" : "0");
zconfig_put (config, "store/db", SELFTEST_DIR_RW "/storedb");
zactor_t *tower = zactor_new (dafka_tower_actor, config);
dafka_producer_args_t args = {"dummy", config};
zactor_t *producer = zactor_new (dafka_producer, &args);
assert (producer);
zactor_destroy (&producer);
zactor_destroy (&tower);
zconfig_destroy (&config);dafka_store -
Please add '@discuss' section in './../src/dafka_store.c'.
This is the class interface:
// Create new dafka_store actor instance.
// @TODO: Describe the purpose of this actor!
//
// zactor_t *dafka_store = zactor_new (dafka_store, NULL);
//
// Destroy dafka_store instance.
//
// zactor_destroy (&dafka_store);
//
// Start dafka_store actor.
//
// zstr_sendx (dafka_store, "START", NULL);
//
// Stop dafka_store actor.
//
// zstr_sendx (dafka_store, "STOP", NULL);
//
// This is the dafka_store constructor as a zactor_fn;
DAFKA_EXPORT void
dafka_store_actor (zsock_t *pipe, void *args);
// Self test of this actor
DAFKA_EXPORT void
dafka_store_test (bool verbose);Please add '@interface' section in './../src/dafka_store.c'.
This is the class self test code:
// ----------------------------------------------------
// Cleanup old test artifacts
// ----------------------------------------------------
if (zsys_file_exists (SELFTEST_DIR_RW "/storedb")) {
zdir_t *store_dir = zdir_new (SELFTEST_DIR_RW "/storedb", NULL);
zdir_remove (store_dir, true);
zdir_destroy (&store_dir);
}
// Simple create/destroy test
zconfig_t *config = zconfig_new ("root", NULL);
zconfig_put (config, "beacon/verbose", verbose ? "1" : "0");
zconfig_put (config, "beacon/sub_address", "inproc://store-tower-sub");
zconfig_put (config, "beacon/pub_address", "inproc://store-tower-pub");
zconfig_put (config, "tower/verbose", verbose ? "1" : "0");
zconfig_put (config, "tower/sub_address", "inproc://store-tower-sub");
zconfig_put (config, "tower/pub_address", "inproc://store-tower-pub");
zconfig_put (config, "store/verbose", verbose ? "1" : "0");
zconfig_put (config, "consumer/verbose", verbose ? "1" : "0");
zconfig_put (config, "producer/verbose", verbose ? "1" : "0");
zconfig_put (config, "store/db", SELFTEST_DIR_RW "/storedb");
zconfig_put (config, "consumer/offset/reset", "earliest");
// Creating the store
zactor_t *tower = zactor_new (dafka_tower_actor, config);
// Creating the publisher
dafka_producer_args_t args = {"TEST", config};
zactor_t *producer = zactor_new (dafka_producer, &args);
// Producing before the store is alive, in order to test fetching between producer and store
dafka_producer_msg_t *p_msg = dafka_producer_msg_new ();
dafka_producer_msg_set_content_str (p_msg, "1");
dafka_producer_msg_send (p_msg, producer);
dafka_producer_msg_set_content_str (p_msg, "2");
dafka_producer_msg_send (p_msg, producer);
// Starting the store
zactor_t *store = zactor_new (dafka_store_actor, config);
zclock_sleep (100);
// Producing another message
dafka_producer_msg_set_content_str (p_msg, "3");
dafka_producer_msg_send (p_msg, producer);
zclock_sleep (100);
// Killing the producer, to make sure the HEADs are coming from the store
zactor_destroy (&producer);
// Starting a consumer and check that consumer recv all 3 messages
zactor_t *consumer = zactor_new (dafka_consumer, config);
dafka_consumer_subscribe (consumer, "TEST");
dafka_consumer_msg_t *c_msg = dafka_consumer_msg_new ();
dafka_consumer_msg_recv (c_msg, consumer);
assert (dafka_consumer_msg_streq (c_msg, "1"));
dafka_consumer_msg_recv (c_msg, consumer);
assert (dafka_consumer_msg_streq (c_msg, "2"));
dafka_consumer_msg_recv (c_msg, consumer);
assert (dafka_consumer_msg_streq (c_msg, "3"));
dafka_consumer_msg_destroy (&c_msg);
zactor_destroy (&consumer);
dafka_producer_msg_destroy (&p_msg);
zactor_destroy (&store);
zactor_destroy (&tower);
zconfig_destroy (&config);
// ----------------------------------------------------
// Cleanup test artifacts
// ----------------------------------------------------
zdir_t *store_dir = zdir_new (SELFTEST_DIR_RW "/storedb", NULL);
zdir_remove (store_dir, true);
zdir_destroy (&store_dir);dafka_tower -
Please add '@discuss' section in './../src/dafka_tower.c'.
This is the class interface:
// Create new dafka_tower actor instance.
// @TODO: Describe the purpose of this actor!
//
// zactor_t *dafka_tower = zactor_new (dafka_tower, NULL);
//
// Destroy dafka_tower instance.
//
// zactor_destroy (&dafka_tower);
//
// Start dafka_tower actor.
//
// zstr_sendx (dafka_tower, "START", NULL);
//
// Stop dafka_tower actor.
//
// zstr_sendx (dafka_tower, "STOP", NULL);
//
// This is the dafka_tower constructor as a zactor_fn;
DAFKA_EXPORT void
dafka_tower_actor (zsock_t *pipe, void *args);
// Self test of this actor
DAFKA_EXPORT void
dafka_tower_test (bool verbose);Please add '@interface' section in './../src/dafka_tower.c'.
This is the class self test code:
// Simple create/destroy test
/*
zactor_t *dafka_tower = zactor_new (dafka_tower_actor, NULL);
assert (dafka_tower);
zactor_destroy (&dafka_tower);
*/Man pages are generated from the class header and source files via the doc/mkman tool, and similar functionality in the gitdown tool (http://github.com/imatix/gitdown). The header file for a class must wrap its interface as follows (example is from include/zclock.h):
// @interface
// Sleep for a number of milliseconds
void
zclock_sleep (int msecs);
// Return current system clock as milliseconds
int64_t
zclock_time (void);
// Self test of this class
int
zclock_test (Bool verbose);
// @end
The source file for a class must provide documentation as follows:
/*
@header
...short explanation of class...
@discuss
...longer discussion of how it works...
@end
*/
The source file for a class then provides the self test example as follows:
// @selftest
int64_t start = zclock_time ();
zclock_sleep (10);
assert ((zclock_time () - start) >= 10);
// @end
The template for man pages is in doc/mkman.
Dafka is developed through a test-driven process that guarantees no memory violations or leaks in the code:
- Modify a class or method.
- Update the test method for that class.
- Run the 'selftest' script, which uses the Valgrind memcheck tool.
- Repeat until perfect.
Don't include system headers in source files. The right place for these is dafka_prelude.h. If you need to check against configured libraries and/or headers, include platform.h in the source before including dafka.h.
Do read your code after you write it and ask, "Can I make this simpler?" We do use a nice minimalist and yet readable style. Learn it, adopt it, use it.
Before opening a pull request read our contribution guidelines. Thanks!
TODO
This documentation was generated from dafka/README.txt using Gitdown
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent