一文彻底读懂高性能消息组件Apache Pulsar

马哥Linux运维

共 39996字,需浏览 80分钟

 ·

2022-01-22 21:22

内容介绍

Pulsar介绍

Pulsar关键特性

Pulsar vs Kafka

Pulsar架构设计

Pulsar消息机制

Pulsar Schema

Pulsar Functions

Pulsar Connectors

Pulsar Deployment

Pulsar Admin

Pulsar Manager

Pulsar Flink

什么是Pulsar?

Apache Pulsar 是 Apache 软件基金会顶级项目,是下一代云原生分布式消息流平台,集消息、存储、轻量化函数式计算为一体,采用计算与存储分离架构设计,支持多租户、持久化存储、多机房跨区域数据复制,具有强一致性、高吞吐以及低延时的高可扩展流数据存储特性。

Pulsar 的关键特性

  • Pulsar 的单个实例原生支持多个集群,可跨机房在集群间无缝地完成消息复制。
  • 极低的发布延迟和端到端延迟。
  • 可无缝扩展到超过 一百万 个 topic。
  • 简单的客户端 API,支持 Java、Go、Python 和 C++。
  • 支持多种 topic 订阅模式(独占订阅、共享订阅、故障转移订阅)。
  • 通过 Apache BookKeeper 提供的持久化消息存储机制保证消息传递 。
  • 由轻量级的 serverless 计算框架 Pulsar Functions 实现流原生的数据处理。
  • 基于 Pulsar Functions 的 serverless connector 框架 Pulsar IO 使得数据更易移入、移出 Apache Pulsar。
  • 分层式存储可在数据陈旧时,将数据从热存储卸载到冷/长期存储(如S3、GCS)中。

Pulsar vs Kafka

下方链接为Pulsar与 Kafka详细对比报告,可自行下载查看

https://streamnative.io/en/blog/tech/2020-07-08-pulsar-vs-kafka-part-1

https://streamnative.io/zh/blog/tech/2020-07-22-pulsar-vs-kafka-part-2

  • 性能与可用性

基准测试(StreamNative)

数据来源

https://mp.weixin.qq.com/s/UZJTOEpzX8foUJv9XMJxOw

https://streamnative.io/en/blog/tech/2020-11-09-benchmark-pulsar-kafka-performance

https://streamnative.io/whitepaper/benchmark-pulsar-vs-kafka

  • 吞吐量(Throughput)

在与 Kafka 的持久性保证相同的情况下,Pulsar 可达到 605 MB/s 的发布和端到端吞吐量(与 Kafka 相同)以及 3.5 GB/s 的 catch-up read 吞吐量(比 Kafka 高 3.5 倍)。Pulsar 的吞吐量不会因分区数量的增加和持久性级别的改变而受到影响,而 Kafka 的吞吐量会因分区数量或持久性级别的改变而受到严重影响。

  • 延迟性(Latency)

在不同的测试实例(包括不同订阅数量、不同主题数量和不同持久性保证)中,Pulsar 的延迟显著低于 Kafka。Pulsar P99 延迟在 5 到 15 毫秒之间。Kafka P99 延迟可能长达数秒,并且会因主题数量、订阅数量和不同持久性保证而受到巨大影响。

  • 功能性

  1. 多语言客户端(C/C++、Python、Java、Go ...)
  2. 管理工具(Pulsar Manager vs Kafka Manager)
  3. 内置流处理Built-In Stream Processing(Pulsar Function vs Kafka Streams)
  4. Rich Integrations (Pulsar Connectors)
  5. Exactly-Once Processing
  6. 日志压缩
  7. 多租户(Pulsar)
  8. 安全管理(Pulsar)

架构设计

Pulsar采用存储和计算分离的软件架构。在消息领域,Pulsar 是第一个将存储计算分离云原生架构落地的开源项目。由于在 Broker 层不存储任何数据,这种架构为用户带来了更高的可用性、更灵活的扩容和管理、避免数据的 reblance 和 catch-up。

在 Apache Pulsar 的分层架构中,服务层 Broker 和存储层 BookKeeper 的每个节点都是对等的。Broker 仅仅负责消息的服务支持,不存储数据。这为服务层和存储层提供了瞬时的节点扩展和无缝的失效恢复。

持久化存储(Persistent storage)

Pulsar 使用 BookKeeper 分布式日志存储数据库作为存储组件,在底层使用日志作为存储模型。

Pulsar 将所有未确认消息(即未处理消息)存储在 BookKeeper 中的多个“bookie”服务器上。

BookKeeper 通过 Quorum Vote 的方式来实现数据的一致性,跟 Master/Slave 模式不同,BookKeeper 中每个节点也是对等的,对一份数据会并发地同时写入指定数目的存储节点。

一个Topic实际上是一个ledgers流。Ledger本身就是一个日志。所以一系列的子日志(Ledgers)组成了一个父日志(Topic)。

Ledgers追加到一个Topic,条目(消息或者一组消息)追加到Ledgers。Ledger一旦关闭是不可变的。Ledger作为最小的删除单元,也就是说我们不能删除单个条目而是去删除整个Ledger。

Ledgers本身也被分解为多个Fragment。Fragment是BookKeeper集群中最小的分布单元。

每个Ledger(由一个或多个Fragment组成)可以跨多个BookKeeper节点(Bookies)进行复制,以实现数据容灾和提升读取性能。每个Fragment都在一组不同的Bookies中复制(存在足够的Bookies)。

conf/bookkeeper.conf

#############################################################################
## Server parameters
#############################################################################
# Directories BookKeeper outputs its write ahead log.
# Could define multi directories to store write head logs, separated by ','.
journalDirectories=/data/appData/pulsar/bookkeeper/journal
#############################################################################
## Ledger storage settings
#############################################################################
# Directory Bookkeeper outputs ledger snapshots
# could define multi directories to store snapshots, separated by ','
ledgerDirectories=/data/appData/pulsar/bookkeeper/ledgers

conf/broker.conf

### --- Managed Ledger --- ###
# Number of bookies to use when creating a ledger
managedLedgerDefaultEnsembleSize=2
# Number of copies to store for each message
managedLedgerDefaultWriteQuorum=2
# Number of guaranteed copies (acks to wait before write is complete)
managedLedgerDefaultAckQuorum=2

元数据存储(Metadata storage)

Pulsar和BookKeeper都使用Apache Zookeeper来存储元数据和监控节点健康状况。

$PULSAR_HOME/bin/pulsar zookeeper-shell
> ls /
[admin, bookies, counters, ledgers, loadbalance, managed-ledgers, namespace, pulsar, schemas, stream, zookeeper]

消息机制

Pulsar 采用发布-订阅(pub-sub)的设计模式 。该设计模式中,producer 发布消息到 topic, Consumer 订阅 topic、处理发布的消息,并在处理完成后发送确认。

一旦创建订阅,即使 consumer 断开连接,Pulsar 仍然可以保存所有消息。在 consumer 确认消息已处理成功后,才会删除消息。

主题(Topic)

逻辑上一个Topic是日志结构,每个消息都在这个日志结构中有一个偏移量。Apache Pulsar使用游标来跟踪偏移量(Cursor Tracking)。

Pulsar 支持两种基本的 topic 类型:持久 topic 与非持久 topic。

{persistent|non-persistent}://tenant/namespace/topic
  • Non-Partitioned topics
$PULSAR_HOME/bin/pulsar-admin topics \
list public/default
$PULSAR_HOME/bin/pulsar-admin topics \
create persistent://public/default/input-seed-avro-topic
$PULSAR_HOME/bin/pulsar-admin topics \
lookup persistent://public/default/input-seed-avro-topic
$PULSAR_HOME/bin/pulsar-admin topics \
delete persistent://public/default/input-seed-avro-topic
$PULSAR_HOME/bin/pulsar-admin topics \
stats persistent://public/default/input-seed-avro-topic
$ curl http://server-101:8080/admin/v2/persistent/public/default/exclamation-input/stats | python -m json.tool

Partitioned topics

$PULSAR_HOME/bin/pulsar-admin topics \
create-partitioned-topic persistent://public/default/output-seed-avro-topic \
--partitions 2
$PULSAR_HOME/bin/pulsar-admin topics \
list-partitioned-topics public/default
$PULSAR_HOME/bin/pulsar-admin topics \
get-partitioned-topic-metadata persistent://public/default/output-seed-avro-topic
$PULSAR_HOME/bin/pulsar-admin topics \
delete-partitioned-topic persistent://public/default/output-seed-avro-topic

消息(Message)

Messages are the basic "unit" of Pulsar.

public interface Message<T{
    Map getProperties();
    boolean hasProperty(String var1);
    String getProperty(String var1);
    byte[] getData();
    getValue();
    MessageId getMessageId();
    long getPublishTime();
    long getEventTime();
    long getSequenceId();
    String getProducerName();
    boolean hasKey();
    String getKey();
    boolean hasBase64EncodedKey();
    byte[] getKeyBytes();
    boolean hasOrderingKey();
    byte[] getOrderingKey();
    String getTopicName();
    Optional getEncryptionCtx();
    int getRedeliveryCount();
    byte[] getSchemaVersion();
    boolean isReplicated();
    String getReplicatedFrom();
}

生产者(Producer)

public void send() throws PulsarClientException {
    final String serviceUrl = "pulsar://server-100:6650";
    // final String serviceUrl = "pulsar://server-101:6650,server-102:6650,server-103:6650";
    // http://pulsar.apache.org/docs/en/client-libraries-java/#client
    final PulsarClient client = PulsarClient.builder()
            .serviceUrl(serviceUrl)
            .connectionTimeout(10000, TimeUnit.MILLISECONDS)
            .build();
    final String topic = "persistent://public/default/topic-sensor-temp";
    // http://pulsar.apache.org/docs/en/client-libraries-java/#configure-producer
    final Producer<byte[]> producer = client.newProducer()
            .producerName("sensor-temp")
            .topic(topic)
            .compressionType(CompressionType.LZ4)
            .enableChunking(true)
            .enableBatching(true)
            .batchingMaxBytes(1024)
            .batchingMaxMessages(10)
            .batchingMaxPublishDelay(10, TimeUnit.MILLISECONDS)
            .blockIfQueueFull(true)
            .maxPendingMessages(512)
            .sendTimeout(1, TimeUnit.SECONDS)
            .create();
    MessageId mid = producer.send("sensor-temp".getBytes());
    System.out.printf("\nmessage with ID %s successfully sent", mid);
    mid = producer.newMessage()
            .key("sensor-temp-key")
            .value("sensor-temp-key".getBytes())
            .property("my-key""my-value")
            .property("my-other-key""my-other-value")
            .send();
    System.out.printf("message-key with ID %s successfully sent", mid);
    producer.close();
    client.close();
}

消费者(Consumer)

public void consume() throws PulsarClientException {
    final String serviceUrl = "pulsar://server-101:6650";
    final String topic = "input-seed-avro-topic";
    final PulsarClient client = PulsarClient.builder()
            .serviceUrl(serviceUrl)
            .enableTcpNoDelay(true)
            .build();
    final Consumer<byte[]> consumer = client
            .newConsumer()
            .consumerName("seed-avro-consumer")
            .subscriptionName("seed-avro-subscription")
            .subscriptionType(SubscriptionType.Exclusive)
            .subscriptionInitialPosition(SubscriptionInitialPosition.Earliest)
            .topic(topic)
            .receiverQueueSize(10)
            .subscribe();
    final AvroSchema schema = AvroSchema.of(SeedEvent.class);
    while (true) {
        try {
            final Message<byte[]> msg = consumer.receive();
            LOG.info("接收消息:[{}] topic:{} mid:{} sid:{} event:{} publish:{} producer:{} key:{} value:{}",
                    Thread.currentThread().getId(),
                    msg.getTopicName(),
                    msg.getMessageId(),
                    msg.getSequenceId(),
                    msg.getEventTime(),
                    msg.getPublishTime(),
                    msg.getProducerName(),
                    msg.getKey(), schema.decode(msg.getValue()));
            try {
                consumer.acknowledge(msg);
            } catch (final PulsarClientException e) {
                consumer.negativeAcknowledge(msg);
                LOG.error("acknowledge:" + e.getLocalizedMessage(), e);
            }
        } catch (final PulsarClientException e) {
            LOG.error("receive:" + e.getLocalizedMessage(), e);
        }
    }
}

订阅(Subscriptions)

消费者通过订阅来消费Topic中的消息。订阅是游标(跟踪偏移量)的逻辑实体,一个Topic可以添加多个订阅。订阅不包含消息的数据,只包含元数据和游标。

每个Subscription都存储一个Cursor。Cursor是日志中的当前偏移量。Subscription将其Cursor存储至BookKeeper的Ledger中。这使Cursor跟踪可以像Topic一样进行扩展。

订阅类型(subscription-type)

  • Exclusive 独享

一个订阅只能有一个消息者消费消息。

  • Failover 灾备

一个订阅同时只有一个消费者,可以有多个备份消费者。一旦主消费者故障则备份消费者接管。不会出现同时有两个活跃的消费者。


  • Shared 共享

一个订阅中同时可以有多个消费者,多个消费者共享Topic中的消息。

  • Key_Shared

有序性保证(Ordering guarantee)

如果对顺序性有要求,可以使用 Exclusive 和 Failover 的订阅模式,这样同一个 Topic 只有一个 Consumer 在消费,可以保证顺序性。

如果使用 Shared 订阅模式,多个 Consumer 可以并发消费同一个 Topic。通过动态增加 Consumer 的数量,可以加速 Topic 的消费,减少消息在服务端的堆积。

KeyShared 模式保证在 Shared 模式下同一个 Key 的消息也会发送到同一个 Consumer,在并发的同时也保证了顺序性。

多主题订阅(Multi-topic subscriptions)

Pattern:

  • persistent://public/default/.*
  • persistent://public/default/foo.*

Reader

public void read() throws IOException {
    final String serviceUrl = "pulsar://server-101:6650";
    final PulsarClient client = PulsarClient.builder()
            .serviceUrl(serviceUrl)
            .build();
    // http://pulsar.apache.org/docs/en/client-libraries-java/#reader
    final Reader<byte[]> reader = client.newReader()
            .topic("my-topic")
            .startMessageId(MessageId.earliest()) // MessageId.latest
            .create();
    while (true
        final Message<byte[]> message = reader.readNext();
        System.out.println(new String(message.getData()));
    }
}

分片主题(Partitioned topics)

消息保留和过期(Message retention and expiry)

如果没有对Topic设置数据保留策略,一旦一个Topic的所有订阅的游标都已经成功消费到一个偏移量时,此偏移量前面的消息就会被自动删除。

如果Topic设置了数据保留策略,已经消费确认的消息超过保留策略阈值(Topic的消息存储大小、Topic中消息保留的时间)后会被删除。

conf/broker.conf

# Default message retention time
# 默认0, 修改为3天=60*24*3
defaultRetentionTimeInMinutes=4320
# Default retention size
# 默认为0, 修改为10G
defaultRetentionSizeInMB=10240
# Default ttl for namespaces if ttl is not already configured at namespace policies. (disable default-ttl with value 0)
ttlDurationDefaultInSeconds=0

retention policy (for a namespace)

$PULSAR_HOME/bin/pulsar-admin namespaces \
get-retention public/default
$ curl -X GET http://server-101:8080/admin/v2/namespaces/public/default/retention | python -m json.tool
$PULSAR_HOME/bin/pulsar-admin namespaces \
set-retention public/default \
--size 1024M \
--time 5m
$ curl -X POST http://server-101:8080/admin/v2/namespaces/public/default/retention \
--header "Content-Type:application/json" \
--data '{
  "retentionTimeInMinutes" : 5,
  "retentionSizeInMB" : 1024
}'

message expiry / message-ttl

$PULSAR_HOME/bin/pulsar-admin namespaces \
get-message-ttl public/default
$ curl -X GET http://server-101:8080/admin/v2/namespaces/public/default/messageTTL
$PULSAR_HOME/bin/pulsar-admin namespaces \
set-message-ttl public/default \
--messageTTL 1800
$ curl -X POST http://server-101:8080/admin/v2/namespaces/public/default/messageTTL \
--header "Content-Type:application/json" \
--data '1800'

Pulsar Schema

Pulsar schema enables you to use language-specific types of data when constructing and handling messages from simple types to more complex application-specific types.

  • 类型安全(序列化和反序列化)
  • Schema 帮助 Pulsar 保留了数据在其他系统中原有的含义

Schema类型(Schema type)

  • Primitive type
Producer producer = client.newProducer(Schema.STRING).create();
producer.newMessage().value("Hello Pulsar!").send();
Consumer consumer = client.newConsumer(Schema.STRING).subscribe();
consumer.receive();
  • Complex type

1. keyvalue key/value pair.

Schema> schema = Schema.KeyValue(
Schema.INT32,
Schema.STRING,
KeyValueEncodingType.SEPARATED
);
// Producer
Producer> producer = client.newProducer(schema)
    .topic(TOPIC)
    .create();
final int key = 100;
final String value = "value-100";
producer.newMessage().value(new KeyValue<>(key, value)).send();
// Consumer
Consumer> consumer = client.newConsumer(schema)
    .topic(TOPIC).subscriptionName(SubscriptionName).subscribe();
Message> msg = consumer.receive();

2.struct AVRO, JSON, and Protobuf.

Producer producer = client.newProducer(Schema.AVRO(User.class)).create();
producer.newMessage().value(User.builder().userName("pulsar-user").userId(1L).build()).send();
Consumer consumer = client.newConsumer(Schema.AVRO(User.class)).subscribe();
User user = consumer.receive();

Schema工作方式(How does schema work)

Producer

Consumer

Schema管理(Schema manual management)

查询Schema

$PULSAR_HOME/bin/pulsar-admin schemas \
get persistent://public/default/spirit-avro-topic
$PULSAR_HOME/bin/pulsar-admin schemas \
get persistent://public/default/spirit-avro-topic \
--version=2

更新Schema

$PULSAR_HOME/bin/pulsar-admin schemas upload \
persistent://public/default/test-topic \
--filename $PULSAR_HOME/connectors/json-schema.json

提取Schema

$PULSAR_HOME/bin/pulsar-admin schemas \
extract persistent://public/default/test-topic  \
--classname com.cloudwise.modal.Packet \
--jar ~/cloudwise-pulsar-1.0.0-RELEASE.jar \
--type json
public void schemaInfo() {
    System.out.println("AvroSchema:" + AvroSchema.of(SeedEvent.class).getSchemaInfo());
    System.out.println("Schema.AVRO:" + Schema.AVRO(SeedEvent.class).getSchemaInfo());
}

删除Schema

$PULSAR_HOME/bin/pulsar-admin schemas \
delete persistent://public/default/spirit-avro-topic

Pulsar Functions

编程模型(Programming model)

开启Functions

  1. conf/bookkeeper.conf
extraServerComponents=org.apache.bookkeeper.stream.server.StreamStorageLifecycleComponent
  1. conf/broker.conf
functionsWorkerEnabled=true
  1. conf/functions_worker.yml
pulsarFunctionsCluster: pulsar-cluster
numFunctionPackageReplicas: 2

窗口(window)

  • windowLengthCount 每个窗口的消息数量
  • slidingIntervalCount 窗口滑动后的消息数量
  • windowLengthDurationMs 窗口时间
  • slidingIntervalDurationMs 窗口滑动后的时间

开窗函数

public class WordCountWindowFunction implements org.apache.pulsar.functions.api.WindowFunction<StringVoid{
    @Override
    public Void process(Collection> inputs, WindowContext context) throws Exception {
        for (Record input : inputs) {
        }
        return null;
    }
}

运行函数

  • 时间,滑动窗口

--user-config '{"windowLengthDurationMs":"60000", "slidingIntervalDurationMs":"1000"}'

  • 时间,滚动窗口

--user-config '{"windowLengthDurationMs":"60000"}'

  • 数量,滑动窗口

--user-config '{"windowLengthCount":"100", "slidingIntervalCount":"10"}'

  • 数量,滚动窗口

--user-config '{"windowLengthCount":"100"}'

Java编程

pom.xml

<dependency>
    <groupId>org.apache.pulsargroupId>
    <artifactId>pulsar-clientartifactId>
    <version>${pulsar.version}version>
dependency>
<dependency>
    <groupId>org.apache.pulsargroupId>
    <artifactId>pulsar-functions-apiartifactId>
    <version>${pulsar.version}version>
dependency>
<dependency>
    <groupId>org.apache.pulsargroupId>
    <artifactId>pulsar-functions-local-runnerartifactId>
    <version>${pulsar.version}version>
dependency>
  1. WordCount
public class WordCountFunction implements org.apache.pulsar.functions.api.Function<StringVoid{
    @Override
    public Void process(String input, Context context) throws Exception {
        Arrays.asList(input.split(" ")).forEach(word -> {
            String counterKey = word.toLowerCase();
            if (context.getCounter(counterKey) == 0) {
                context.putState(counterKey, ByteBuffer.wrap(ByteUtils.from(100)));
            }
            context.incrCounter(counterKey, 1);
        });
        return null;
    }
}
$ $PULSAR_HOME/bin/pulsar-admin functions create \
--broker-service-url pulsar://server-101:6650 \
--jar target/cloudwise-pulsar-functions-with-dependencies.jar \
--classname com.cloudwise.quickstart.pulsar.functions.WordCountFunction \
--tenant public \
--namespace default \
--name word-count-function \
--inputs persistent://public/default/sentences \
--output persistent://public/default/wordcount
  1. 动态路由
/**
 * 基本思路是检查每条消息的内容,根据消息内容将消息路由到不同目的地。
 */

public class RoutingFunction implements org.apache.pulsar.functions.api.Function<StringString{
    @Override
    public String process(String input, Context context) throws Exception {
        String regex = context.getUserConfigValue("regex").toString();
        String matchedTopic = context.getUserConfigValue("matched-topic").toString();
        String unmatchedTopic = context.getUserConfigValue("unmatched-topic").toString();
        Pattern pattern = Pattern.compile(regex);
        Matcher matcher = pattern.matcher(input);
        if (matcher.matches()) {
            context.newOutputMessage(matchedTopic, Schema.STRING).value(input).send();
        } else {
            context.newOutputMessage(unmatchedTopic, Schema.STRING).value(input).send();
        }
        return null;
    }
}
  1. log-topic
public class LoggingFunction implements org.apache.pulsar.functions.api.Function<StringVoid{
    @Override
    public Void process(String s, Context context) throws Exception {
        Logger LOG = context.getLogger();
        String messageId = context.getFunctionId();
        if (s.contains("danger")) {
            LOG.warn("A warning was received in message {}", messageId);
        } else {
            LOG.info("Message {} received\nContent: {}", messageId, s);
        }
        return null;
    }
}
$ $PULSAR_HOME/bin/pulsar-admin functions create \
--jar cloudwise-pulsar-functions-1.0.0.jar \
--classname com.cloudwise.quickstart.pulsar.functions.LoggingFunction \
--log-topic persistent://public/default/logging-function-logs
  1. user-config
public class UserConfigFunction implements org.apache.pulsar.functions.api.Function<StringVoid{
    @Override
    public Void process(String s, Context context) throws Exception {
        Logger log = context.getLogger();
        Optional value = context.getUserConfigValue("word-of-the-day");
        if (value.isPresent()) {
            log.info("The word of the day is {}", value);
        } else {
            log.warn("No word of the day provided");
        }
        return null;
    }
}
$ $PULSAR_HOME/bin/pulsar-admin functions create \
--broker-service-url pulsar://server-101:6650 \
--jar target/cloudwise-pulsar-functions-with-dependencies.jar \
--classname com.cloudwise.quickstart.pulsar.functions.UserConfigFunction \
--tenant public \
--namespace default \
--name word-count-function \
--inputs persistent://public/default/userconfig \
--user-config '{"word-of-the-day":"verdure"}'

Pulsar Connectors

img

消息处理(Processing guarantee)

  • at-most-once
  • at-least-once
  • effectively-once

操作流程(JDBC sink)

  1. Add a configuration file.
  2. Create a schema.
  3. Upload a schema to a topic.
  4. Create a JDBC sink
  5. Stop a JDBC sink
  6. Restart a JDBC sink
  7. Update a JDBC sink

内建连接器(Built-in connector)

Source connector

  • Canal
  • File
  • Flume
  • Kafka
  • RabbitMQ

Sink connector

  • ElasticSearch/Solr
  • Flume
  • HBase
  • HDFS2/HDFS3
  • InfluxDB
  • JDBC ClickHouse/MariaDB/PostgreSQL
  • Kafka
  • MongoDB
  • RabbitMQ
  • Redis

ClickHouse Sink

  1. 创建表
CREATE DATABASE IF NOT EXISTS monitor;
CREATE TABLE IF NOT EXISTS monitor.pulsar_clickhouse_jdbc_sink
(
    id   UInt32,
    name String
ENGINE = TinyLog;
INSERT INTO monitor.pulsar_clickhouse_jdbc_sink (idname)
VALUES (1'tmp');
SELECT *
FROM monitor.pulsar_clickhouse_jdbc_sink;
  1. 创建配置
$ vi $PULSAR_HOME/connectors/pulsar-clickhouse-jdbc-sink.yaml
{
    "userName""sysop",
    "password""123456",
    "jdbcUrl""jdbc:clickhouse://server-101:8123/monitor",
    "tableName""pulsar_clickhouse_jdbc_sink"
}
  1. 创建schema
$ vi $PULSAR_HOME/connectors/json-schema.json
{
  "name""",
  "schema": {
    "type""record",
    "name""SeedEvent",
    "namespace""com.cloudwise.quickstart.model",
    "fields": [
      {
        "name""id",
        "type": [
          "null",
          "int"
        ]
      },
      {
        "name""name",
        "type": [
          "null",
          "string"
        ]
      }
    ]
  },
  "type""JSON",
  "properties": {
    "__alwaysAllowNull""true",
    "__jsr310ConversionEnabled""false"
  }
}
  1. 上传schema
$PULSAR_HOME/bin/pulsar-admin schemas upload \
pulsar-postgres-jdbc-sink-topic \
-f $PULSAR_HOME/connectors/json-schema.json
  1. 运行
$PULSAR_HOME/bin/pulsar-admin sinks create \
--tenant public \
--namespace default \
--name pulsar-clickhouse-jdbc-sink \
--inputs pulsar-clickhouse-jdbc-sink-topic \
--sink-config-file $PULSAR_HOME/connectors/pulsar-clickhouse-jdbc-sink.yaml \
--archive $PULSAR_HOME/connectors/pulsar-io-jdbc-clickhouse-2.6.2.nar \
--processing-guarantees EFFECTIVELY_ONCE \
--parallelism 1

Pulsar Deployment

目录结构

/opt/pulsar-2.6.2

├── bin

│ ├── bookkeeper

│ ├── function-localrunner

│ ├── proto

│ ├── pulsar

│ ├── pulsar-admin

│ ├── pulsar-admin-common.sh

│ ├── pulsar-client

│ ├── pulsar-daemon

│ ├── pulsar-managed-ledger-admin

│ └── pulsar-perf

├── conf

│ ├── bkenv.sh

│ ├── bookkeeper.conf

│ ├── broker.conf

│ ├── client.conf

│ ├── discovery.conf

│ ├── filesystem_offload_core_site.xml

│ ├── functions-logging

│ ├── functions_worker.yml

│ ├── global_zookeeper.conf

│ ├── log4j2-scripts

│ ├── log4j2.yaml

│ ├── presto

│ ├── proxy.conf

│ ├── pulsar_env.sh

│ ├── pulsar_tools_env.sh

│ ├── schema_example.conf

│ ├── standalone.conf

│ ├── websocket.conf

│ └── zookeeper.conf

├── examples

│ ├── api-examples.jar

│ ├── example-function-config.yaml

│ ├── example-window-function-config.yaml

│ └── python-examples

├── instances

│ ├── deps

│ ├── java-instance.jar

│ └── python-instance

├── lib

│ └── presto

├── LICENSE

├── licenses

├── NOTICE

└── README

单机(Standalone)

# 前台启动
$PULSAR_HOME/bin/pulsar standalone
# 后台启动
$PULSAR_HOME/bin/pulsar-daemon start standalone
$ jps | grep -v Jps
1873 PulsarStandaloneStarter
# 后台停止
$PULSAR_HOME/bin/pulsar-daemon stop standalone -force

集群(Cluster)

  1. 部署ZooKeeper集群
  2. 初始化集群元信息
  3. 部署BookKeeper集群
  4. 部署一个或多个PulsarBroker

客户端(Client)

# consumer
$PULSAR_HOME/bin/pulsar-client consume \
persistent://public/default/seed-avro-topic \
--subscription-name cli-pack-avro-subscription \
--subscription-type Exclusive \
--subscription-position Latest \
--num-messages 0
# producer
$PULSAR_HOME/bin/pulsar-client produce \
persistent://public/default/seed-avro-topic \
--num-produce 100 \
--messages "Hello Pulsar" \
--separator ","

Pulsar Admin

API

  • pulsar-admin
  • REST API

源码:apache-pulsar-2.6.2-src/pulsar-broker/src/main/java/org/apache/pulsar/broker/admin/v2/*.java

Bookies.java Namespaces.java Tenants.java

BrokerStats.java NonPersistentTopics.java Worker.java

Brokers.java PersistentTopics.java WorkerStats.java

Clusters.java ResourceQuotas.java

Functions.java SchemasResource.java

  • Java admin client
public void createNonPartitionedTopic() throws PulsarClientException {
    final String serviceHttpUrl = "http://10.2.2.26:8080";
    final PulsarAdmin admin = PulsarAdmin.builder()
            .serviceHttpUrl(serviceHttpUrl)
            .build();
    try {
        final String namespace = "public/monitor";
        List topics = admin.topics().getList(namespace);
        topics.forEach(t -> System.err.println("before topic:" + t));
        // 以下几种写法是等效的
        // final String topic = "input-3-seed-avro-topic";
        // final String topic = "public/monitor/input-seed-avro-topic";
        final String topic = "persistent://public/default/input-5-seed-avro-topic";
        if (topics.indexOf(topic) == -1) {
            admin.topics().createNonPartitionedTopic(topic);
            admin.schemas().createSchema(topic,
                    AvroSchema.of(SeedEvent.class).getSchemaInfo());
        }
        topics = admin.topics().getList(namespace);
        topics.forEach(t -> System.err.println("after topic:" + t));
        System.err.println("schema:" + admin.schemas().getSchemaInfo(topic));
    } catch (final PulsarAdminException e) {
        e.printStackTrace();
    }
    admin.close();
}

Manage Pulsar

  • Clusters
$PULSAR_HOME/bin/pulsar-admin clusters
  • Tenants
$PULSAR_HOME/bin/pulsar-admin tenants
  • Brokers
$PULSAR_HOME/bin/pulsar-admin brokers
  • Namespaces
$PULSAR_HOME/bin/pulsar-admin namespaces
  • Permissions
  • Persistent topics
  • Non-Persistent topics
  • Partitioned topics
  • Non-Partitioned topics
$PULSAR_HOME/bin/pulsar-admin topics
  • Schemas
$PULSAR_HOME/bin/pulsar-admin schemas
  • Functions
$PULSAR_HOME/bin/pulsar-admin functions

Pulsar Manager

http://pulsar.apache.org/docs/zh-CN/administration-pulsar-manager/

https://github.com/apache/pulsar-manager

WebUI

http://localhost:7750/ui/index.html

username/password: admin/123456

  • Environments
  • Management
  • Clusters
  • Tenants
  • Namespaces
  • Topics
  • Tokents

Pulsar Flink

https://github.com/streamnative/pulsar-flink

https://dl.bintray.com/streamnative/maven/io/streamnative/connectors/

pom.xml

<properties>
    <project.build.sourceEncoding>UTF-8project.build.sourceEncoding>
    <project.reporting.outputEncoding>UTF-8project.reporting.outputEncoding>
    <java.version>1.8java.version>
    <flink.version>1.11.2flink.version>
    <scala.binary.version>2.12scala.binary.version>
    <maven.compiler.source>${java.version}maven.compiler.source>
    <maven.compiler.target>${java.version}maven.compiler.target>
properties>
<repositories>
    <repository>
        <id>centralid>
        <layout>defaultlayout>
        <url>https://repo1.maven.org/maven2url>
    repository>
    <repository>
        <id>bintrayid>
        <name>bintrayname>
        <url>https://dl.bintray.com/streamnative/mavenurl>
    repository>
repositories>
<dependencies>
    <dependency>
        <groupId>org.apache.flinkgroupId>
        <artifactId>flink-clients_${scala.binary.version}artifactId>
        <version>${flink.version}version>
        <scope>providedscope>
    dependency>
    <dependency>
        <groupId>org.apache.flinkgroupId>
        <artifactId>flink-javaartifactId>
        <version>${flink.version}version>
        <scope>providedscope>
    dependency>
    <dependency>
        <groupId>org.apache.flinkgroupId>
        <artifactId>flink-streaming-java_${scala.binary.version}artifactId>
        <version>${flink.version}version>
        <scope>providedscope>
    dependency>
    <dependency>
        <groupId>org.apache.flinkgroupId>
        <artifactId>flink-table-planner_${scala.binary.version}artifactId>
        <version>${flink.version}version>
        <scope>providedscope>
    dependency>
    
    <dependency>
        <groupId>org.apache.flinkgroupId>
        <artifactId>flink-statebackend-rocksdb_${scala.binary.version}artifactId>
        <version>${flink.version}version>
        <scope>providedscope>
    dependency>
    
    <dependency>
        <groupId>io.streamnative.connectorsgroupId>
        <artifactId>pulsar-flink-${scala.binary.version}-${flink.version}artifactId>
        <version>2.5.4.1version>
    dependency>
    
    <dependency>
        <groupId>org.apache.flinkgroupId>
        <artifactId>flink-avroartifactId>
        <version>${flink.version}version>
    dependency>
    <dependency>
        <groupId>org.apache.flinkgroupId>
        <artifactId>flink-jsonartifactId>
        <version>${flink.version}version>
    dependency>
    <dependency>
        <groupId>org.slf4jgroupId>
        <artifactId>slf4j-log4j12artifactId>
        <version>1.7.7version>
        <scope>runtimescope>
    dependency>
    <dependency>
        <groupId>log4jgroupId>
        <artifactId>log4jartifactId>
        <version>1.2.17version>
        <scope>runtimescope>
    dependency>
    <dependency>
        <groupId>junitgroupId>
        <artifactId>junitartifactId>
        <version>4.12version>
        <scope>testscope>
    dependency>
dependencies>

FlinkPulsarSink

public class PulsarSinkJob {
    private static final Logger LOG = LoggerFactory.getLogger(PulsarSinkJob.class);
    public static SourceFunction getSeedSource() {
        final int interval = 5000;
        return new PeriodicEventSource<>(
                Integer.MAX_VALUE, interval, new PeriodicEventSource.Creator() {
            private static final long serialVersionUID = 1L;
            @Override
            public Collection build(long i) {
                return Arrays.stream(new String[]{"TEM-A-01""HUM-A-01""PRS-A-01"})
                        .map(code -> {
                            final SeedEvent event = new SeedEvent(
                                    Instant.now().toEpochMilli(), code, Long.toString(i));
                            LOG.info("创建消息:[{}] {}", Thread.currentThread().getId(), event);
                            return event;
                        })
                        .collect(Collectors.toList());
            }
            @Override
            public Class clazz() {
                return SeedEvent.class;
            }
        });
    }
    public static FlinkPulsarSink getPulsarSink(ParameterTool params) {
        // String adminUrl = "http://server-101:8080,server-102:8080,server-103:8080";
        final String serviceUrl = params.get("serviceUrl""pulsar://10.2.2.26:6650");
        final String adminUrl = params.get("adminUrl""http://10.2.2.26:8080");
        final String outputTopic = params.get("topic""output-seed-avro-topic");
        final String authPlugin = "org.apache.pulsar.client.impl.auth.AuthenticationToken";
        final String authParams = params.get("authParams");
        final Properties props = new Properties();
        props.setProperty(PulsarOptions.FLUSH_ON_CHECKPOINT_OPTION_KEY, "true");
        props.setProperty(PulsarOptions.PARTITION_DISCOVERY_INTERVAL_MS_OPTION_KEY, "5000");
        final ClientConfigurationData clientConf = new ClientConfigurationData();
        clientConf.setServiceUrl(serviceUrl);
        clientConf.setConnectionTimeoutMs(6000);
        clientConf.setUseTcpNoDelay(true);
        if (!StringUtils.isNullOrWhitespaceOnly(authParams)) {
            clientConf.setUseTls(true);
            clientConf.setAuthPluginClassName(authPlugin);
            clientConf.setAuthParams(authParams);
        }
        final TopicKeyExtractor topicKeyExtractor = new TopicKeyExtractor() {
            private static final long serialVersionUID = 1L;
            @Override
            public byte[] serializeKey(SeedEvent element) {
                LOG.info("serializeKey:[{}] {}", Thread.currentThread().getId(), element);
                return element.getCode().getBytes();
            }
            @Override
            public String getTopic(SeedEvent element) {
                return null;
            }
        };
        final FlinkPulsarSink sink = new FlinkPulsarSink<>(
                adminUrl, Optional.of(outputTopic), clientConf, props, topicKeyExtractor, SeedEvent.class);
        return sink;
    }
    @SuppressWarnings("deprecation")
    public static void main(String[] args) {
        final ParameterTool params = ParameterTool.fromArgs(args);
        final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        env.setStreamTimeCharacteristic(TimeCharacteristic.ProcessingTime);
        env.setRestartStrategy(RestartStrategies.fixedDelayRestart(410000));
        env.setStateBackend(new RocksDBStateBackend(new FsStateBackend("file:///tmp/checkpoint/")));
        env.enableCheckpointing(5000, CheckpointingMode.EXACTLY_ONCE);
        // Job取消和故障时会保留Checkpoint数据, 以便根据实际需要恢复到指定的Checkpoint
        env.getCheckpointConfig().enableExternalizedCheckpoints(
                CheckpointConfig.ExternalizedCheckpointCleanup.RETAIN_ON_CANCELLATION);
        // 确保Checkpoint之间有至少500ms的间隔(Checkpoint最小间隔)
        env.getCheckpointConfig().setMinPauseBetweenCheckpoints(500);
        // 检查点必须在一分钟内完成, 否则被丢弃(Checkpoint的超时时间)
        env.getCheckpointConfig().setCheckpointTimeout(60000);
        // 同一时间只允许进行一个检查点
        env.getCheckpointConfig().setMaxConcurrentCheckpoints(1);
        env.getConfig().setGlobalJobParameters(params);
        // DataStream stream = env.fromCollection(getSeedEvents()).name("Collection");
        final DataStream stream = env.addSource(getSeedSource()).name("SourceFunction");
        final DataStream result = stream
                .keyBy(new KeySelector() {
                    private static final long serialVersionUID = 1L;
                    @Override
                    public String getKey(SeedEvent value) throws Exception {
                        return value.getCode();
                    }
                })
                .process(new KeyedProcessFunction() {
                    private static final long serialVersionUID = 1L;
                    private Map infos;
                    private transient ListState state;
                    @Override
                    public void open(Configuration parameters) throws Exception {
                        LOG.info("open...");
                        this.state = getRuntimeContext().getListState(
                                new ListStateDescriptor<>("state", String.class));
                        this.infos = new HashMap<>();

                        this.infos.put("open", LocalDateTime.now().toString());
                    }
                    @Override
                    public void close() throws Exception {
                        LOG.info("close...");
                    }
                    @Override
                    public void processElement(SeedEvent value,
                                               KeyedProcessFunction.Context ctx, Collector out)

                            throws Exception 
{
                        LOG.info("processElement...");
                        final StringBuffer buffer = new StringBuffer();
                        this.state.get().forEach(t -> buffer.append(t));
                        LOG.info("CurrentKey:{} Input:{} State:{} Infos:{}",
                                ctx.getCurrentKey(), value, buffer, this.infos);
                        value.setPayload("[Prev]" + value.getPayload());
                        this.state.clear();
                        this.state.add(value.toString());
                        out.collect(value);
                    }
                })
                .setParallelism(1);
        result
                .print()
                .setParallelism(1);
        result
                .addSink(getPulsarSink(params))
                .name("FlinkPulsarSink")
                .setParallelism(2);
        LOG.info("ExecutionPlan:{}", env.getExecutionPlan());
        try {
            env.execute("PulsarSinkJob");
        } catch (final Exception e) {
            e.printStackTrace();
        }
    }
}

FlinkPulsarSource

public class PulsarSourceJob {
    private static final Logger LOG = LoggerFactory.getLogger(PulsarSourceJob.class);
    public static FlinkPulsarSource getPulsarSource(ParameterTool params) {
        // String serviceUrl = "pulsar://server-101:6650,server-102:6650,server-103:6650";
        // String adminUrl = "http://server-101:8080,server-102:8080,server-103:8080";
        // final String serviceUrl = params.get("serviceUrl", "pulsar://server-101:6650");
        // final String adminUrl = params.get("adminUrl", "http://server-101:8080");
        final String serviceUrl = params.get("serviceUrl""pulsar://10.2.2.26:6650");
        final String adminUrl = params.get("adminUrl""http://10.2.2.26:8080");
        // final String inputTopic = params.get("topic", "input-1-seed-avro-topic");
        // final String subscription = params.get("subscription", "seed-subscription");
        final String inputTopics = params.get("topic""persistent://public/yang11/zlp.gjsjbz.gjbzcd3");
        // final String inputTopics = params.get("topic", "public/monitor/input-0-seed-avro-topic");
        final String subscription = params.get("subscription""mutil-seed-subscription");
        // final String inputTopicPatten = params.get("topicPatten", "input-1-seed-avro-topic");
        // final String subscription = params.get("subscription", "patten-seed-subscription");
        final String authPlugin = "org.apache.pulsar.client.impl.auth.AuthenticationToken";
        final String authParams = params.get("authParams");
        final Properties props = new Properties();
        // http://pulsar.apache.org/docs/en/client-libraries-java/#reader
        props.setProperty(PulsarOptions.PULSAR_READER_OPTION_KEY_PREFIX + "receiverQueueSize""2000");
        // props.setProperty(PulsarOptions.TOPIC_SINGLE_OPTION_KEY, inputTopic);
        props.setProperty(PulsarOptions.TOPIC_MULTI_OPTION_KEY, inputTopics);
        // props.setProperty(PulsarOptions.TOPIC_PATTERN_OPTION_KEY, inputTopicPatten);
        // org.apache.flink.streaming.connectors.pulsar.internal.SourceSinkUtils.getPartitionDiscoveryIntervalInMillis
        props.setProperty(PulsarOptions.PARTITION_DISCOVERY_INTERVAL_MS_OPTION_KEY, "5000"); // 自动发现topic时间间隔,默认-1
        // org.apache.flink.streaming.connectors.pulsar.internal.SourceSinkUtils.getClientCacheSize
        props.setProperty(PulsarOptions.CLIENT_CACHE_SIZE_OPTION_KEY, "5");
        // org.apache.flink.streaming.connectors.pulsar.internal.SourceSinkUtils.flushOnCheckpoint
        props.setProperty(PulsarOptions.FLUSH_ON_CHECKPOINT_OPTION_KEY, "true");
        // org.apache.flink.streaming.connectors.pulsar.internal.SourceSinkUtils.failOnWrite
        props.setProperty(PulsarOptions.FAIL_ON_WRITE_OPTION_KEY, "false");
        // org.apache.flink.streaming.connectors.pulsar.internal.SourceSinkUtils.getPollTimeoutMs
        props.setProperty(PulsarOptions.POLL_TIMEOUT_MS_OPTION_KEY, "120000");
        // org.apache.flink.streaming.connectors.pulsar.internal.PulsarFetcher
        props.setProperty(PulsarOptions.SUBSCRIPTION_ROLE_OPTION_KEY, subscription);
        // org.apache.flink.streaming.connectors.pulsar.internal.SourceSinkUtils.getCommitMaxRetries
        props.setProperty(PulsarOptions.COMMIT_MAX_RETRIES, "3");
        // org.apache.flink.streaming.connectors.pulsar.internal.PulsarFetcher.PulsarFetcher
        props.setProperty(PulsarOptions.FAIL_ON_DATA_LOSS_OPTION_KEY, "false");
        final ClientConfigurationData clientConf = new ClientConfigurationData();
        clientConf.setServiceUrl(serviceUrl);
        clientConf.setConnectionTimeoutMs(6000);
        if (!StringUtils.isNullOrWhitespaceOnly(authParams)) {
            clientConf.setUseTls(true);
            clientConf.setAuthPluginClassName(authPlugin);
            clientConf.setAuthParams(authParams);
        }
        PulsarDeserializationSchema deserializer = null;
        deserializer = new PulsarDeserializationSchemaWrapper<>(AvroDeser.of(SeedEvent.class));
        deserializer = new PulsarDeserializationSchema() {
            private static final long serialVersionUID = 1L;
            private final DeserializationSchema schema = AvroDeser.of(SeedEvent.class);
            public void open(DeserializationSchema.InitializationContext context) throws Exception {
                this.schema.open(context);
            }
            @Override
            public TypeInformation getProducedType() {
                return this.schema.getProducedType();
            }
            @Override
            public boolean isEndOfStream(SeedEvent nextElement) {
                return this.schema.isEndOfStream(nextElement);
            }
            @Override
            public SeedEvent deserialize(@SuppressWarnings("rawtypes") Message message) throws IOException {
                LOG.info("{}"new String(message.getData()));
                final SeedEvent value = new SeedEvent();
                // final SeedEvent value = this.schema.deserialize(message.getData());
                LOG.info("接收消息:[{}] topic:{} mid:{} sid:{} event:{} publish:{} producer:{} key:{} value:{}",
                        Thread.currentThread().getId(),
                        message.getTopicName(),
                        message.getMessageId(),
                        message.getSequenceId(),
                        message.getEventTime(),
                        message.getPublishTime(),
                        message.getProducerName(),
                        message.getKey(), value);
                return value;
            }
        };
        final FlinkPulsarSource source = new FlinkPulsarSource<>(
                adminUrl, clientConf, deserializer, props);
        source.setStartFromEarliest();
        // source.setStartFromSubscription(subscription);
        // source.setStartFromLatest();
        return source;
    }
    @SuppressWarnings("deprecation")
    public static void main(String[] args) {
        final ParameterTool params = ParameterTool.fromArgs(args);
        final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        env.setStreamTimeCharacteristic(TimeCharacteristic.ProcessingTime);
        env.setStateBackend(new RocksDBStateBackend(new FsStateBackend("file:///tmp/checkpoint/")));
        env.setRestartStrategy(RestartStrategies.fixedDelayRestart(410000));
        // 必须开启Checkpoint, 才能从上一次未消费处开始消费, 否则从头开始消费
        env.enableCheckpointing(5000, CheckpointingMode.EXACTLY_ONCE);
        env.getConfig().setGlobalJobParameters(params);
        env.setParallelism(1);
        final DataStream stream = env
                .addSource(getPulsarSource(params))
                .name("FlinkPulsarSource")
                .uid("PulsarSource")
                .setParallelism(1)
                .map(new MapFunction() {
                    private static final long serialVersionUID = 1L;
                    @Override
                    public String map(SeedEvent value) throws Exception {
                        return "[SourceJob]" + value;
                    }
                });
        stream
                .print()
                .name("[Print]")
                .uid("PrintSink")
                .setParallelism(1);
        try {
            env.execute("PulsarSourceJob");
        } catch (final Exception e) {
            e.printStackTrace();
        }
    }
}

讲师介绍:Larry Zhang (张磊) ,云智慧服务工程部-服务效能二部运维开发工程师,致力于云智慧开源项目运维管理平台(OMP)的产品研发,支撑内部交付体系快速部署监控产品,拥有丰富的运维管理平台和PaaS平台设计开发经验。

(版权归原作者所有,侵删)


点击下方“阅读原文”查看更多

浏览 35
点赞
评论
收藏
分享

手机扫一扫分享

分享
举报
评论
图片
表情
推荐
点赞
评论
收藏
分享

手机扫一扫分享

分享
举报