【问题标题】:Adding custom dependencies for a Plugin in a Flink cluster为 Flink 集群中的插件添加自定义依赖项
【发布时间】:2021-03-17 20:49:09
【问题描述】:

我有一个 Flink 会话集群(作业管理器 + 任务管理器),版本 1.11.1,配置了 log4j-console.properties 以包含 Kafka 附加程序。此外,在作业管理器和任务管理器中,我都启用了 flink-s3-fs-hadoop 内置插件。

我已将 kafka-clients jar 添加到 flink/lib 目录,这是容器运行所必需的。但是在实例化 S3 插件(并初始化记录器)时,我仍然收到以下类加载错误。

原因:org.apache.kafka.common.config.ConfigException:无效值 org.apache.kafka.common.serialization.ByteArraySerializer 用于配置 key.serializer:类 org.apache.kafka.common.serialization .ByteArraySerializer 找不到。

(底部的完整堆栈跟踪)

据我了解,插件有一个专用的动态类加载,它与系统类加载是分开的。因此,我在 flink-conf.yaml 文件中添加了如下配置:

classloader.parent-first-patterns.additional: org.apache.kafka
classloader.resolve-order: parent-first

但错误仍然出现。

在调试时,我没有看到插件类加载器的“allowedFlinkPackages”中添加了额外的模式。

我在这里错过了什么?

java.util.ServiceConfigurationError: org.apache.flink.core.fs.FileSystemFactory: Provider org.apache.flink.fs.s3hadoop.S3FileSystemFactory could not be instantiated
        at java.util.ServiceLoader.fail(Unknown Source) ~[?:?]
        at java.util.ServiceLoader$ProviderImpl.newInstance(Unknown Source) ~[?:?]
        at java.util.ServiceLoader$ProviderImpl.get(Unknown Source) ~[?:?]
        at java.util.ServiceLoader$3.next(Unknown Source) ~[?:?]
        at org.apache.flink.core.plugin.PluginLoader$ContextClassLoaderSettingIterator.next(PluginLoader.java:103) ~[flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.shaded.guava18.com.google.common.collect.Iterators$5.next(Iterators.java:558) ~[flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.shaded.guava18.com.google.common.collect.TransformedIterator.next(TransformedIterator.java:48) ~[flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.core.fs.FileSystem.addAllFactoriesToList(FileSystem.java:1068) [flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.core.fs.FileSystem.loadFileSystemFactories(FileSystem.java:1050) [flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.core.fs.FileSystem.initialize(FileSystem.java:325) [flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.runtime.taskexecutor.TaskManagerRunner.runTaskManagerSecurely(TaskManagerRunner.java:315) [flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.runtime.taskexecutor.TaskManagerRunner.main(TaskManagerRunner.java:297) [flink-dist_2.11-1.11.1.jar:1.11.1]
Caused by: java.lang.ExceptionInInitializerError
        at jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method) ~[?:?]
        at jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance(Unknown Source) ~[?:?]
        at jdk.internal.reflect.DelegatingConstructorAccessorImpl.newInstance(Unknown Source) ~[?:?]
        at java.lang.reflect.Constructor.newInstance(Unknown Source) ~[?:?]
        ... 11 more
Caused by: org.apache.kafka.common.config.ConfigException: Invalid value org.apache.kafka.common.serialization.ByteArraySerializer for configuration key.serializer: Class org.apache.kafka.common.serialization.ByteArraySerializer could not be found.
        at org.apache.kafka.common.config.ConfigDef.parseType(ConfigDef.java:728) ~[kafka-clients-2.5.0.jar:?]
        at org.apache.kafka.common.config.ConfigDef.parseValue(ConfigDef.java:474) ~[kafka-clients-2.5.0.jar:?]
        at org.apache.kafka.common.config.ConfigDef.parse(ConfigDef.java:467) ~[kafka-clients-2.5.0.jar:?]
        at org.apache.kafka.common.config.AbstractConfig.<init>(AbstractConfig.java:108) ~[kafka-clients-2.5.0.jar:?]
        at org.apache.kafka.common.config.AbstractConfig.<init>(AbstractConfig.java:129) ~[kafka-clients-2.5.0.jar:?]
        at org.apache.kafka.clients.producer.ProducerConfig.<init>(ProducerConfig.java:481) ~[kafka-clients-2.5.0.jar:?]
        at org.apache.kafka.clients.producer.KafkaProducer.<init>(KafkaProducer.java:326) ~[kafka-clients-2.5.0.jar:?]
        at org.apache.kafka.clients.producer.KafkaProducer.<init>(KafkaProducer.java:298) ~[kafka-clients-2.5.0.jar:?]
        at org.apache.logging.log4j.core.appender.mom.kafka.DefaultKafkaProducerFactory.newKafkaProducer(DefaultKafkaProducerFactory.java:40) ~[log4j-core-2.12.1.jar:2.12.1]
        at org.apache.logging.log4j.core.appender.mom.kafka.KafkaManager.startup(KafkaManager.java:136) ~[log4j-core-2.12.1.jar:2.12.1]
        at org.apache.logging.log4j.core.appender.mom.kafka.KafkaAppender.start(KafkaAppender.java:164) ~[log4j-core-2.12.1.jar:2.12.1]
        at org.apache.logging.log4j.core.config.AbstractConfiguration.start(AbstractConfiguration.java:304) ~[log4j-core-2.12.1.jar:2.12.1]
        at org.apache.logging.log4j.core.LoggerContext.setConfiguration(LoggerContext.java:579) ~[log4j-core-2.12.1.jar:2.12.1]
        at org.apache.logging.log4j.core.LoggerContext.reconfigure(LoggerContext.java:651) ~[log4j-core-2.12.1.jar:2.12.1]
        at org.apache.logging.log4j.core.LoggerContext.reconfigure(LoggerContext.java:668) ~[log4j-core-2.12.1.jar:2.12.1]
        at org.apache.logging.log4j.core.LoggerContext.start(LoggerContext.java:253) ~[log4j-core-2.12.1.jar:2.12.1]
        at org.apache.logging.log4j.core.impl.Log4jContextFactory.getContext(Log4jContextFactory.java:153) ~[log4j-core-2.12.1.jar:2.12.1]
        at org.apache.logging.log4j.core.impl.Log4jContextFactory.getContext(Log4jContextFactory.java:45) ~[log4j-core-2.12.1.jar:2.12.1]
        at org.apache.logging.log4j.LogManager.getContext(LogManager.java:194) ~[log4j-api-2.12.1.jar:2.12.1]
        at org.apache.logging.log4j.spi.AbstractLoggerAdapter.getContext(AbstractLoggerAdapter.java:138) ~[log4j-api-2.12.1.jar:2.12.1]
        at org.apache.logging.slf4j.Log4jLoggerFactory.getContext(Log4jLoggerFactory.java:45) ~[log4j-slf4j-impl-2.12.1.jar:2.12.1]
        at org.apache.logging.log4j.spi.AbstractLoggerAdapter.getLogger(AbstractLoggerAdapter.java:48) ~[log4j-api-2.12.1.jar:2.12.1]
        at org.apache.logging.slf4j.Log4jLoggerFactory.getLogger(Log4jLoggerFactory.java:30) ~[log4j-slf4j-impl-2.12.1.jar:2.12.1]
        at org.slf4j.LoggerFactory.getLogger(LoggerFactory.java:329) ~[flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.slf4j.LoggerFactory.getLogger(LoggerFactory.java:349) ~[flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.fs.s3.common.AbstractS3FileSystemFactory.<clinit>(AbstractS3FileSystemFactory.java:88) ~[?:?]
        at jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method) ~[?:?]
        at jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance(Unknown Source) ~[?:?]
        at jdk.internal.reflect.DelegatingConstructorAccessorImpl.newInstance(Unknown Source) ~[?:?]
        at java.lang.reflect.Constructor.newInstance(Unknown Source) ~[?:?]
        ... 11 more
[2020-12-06T09:15:45,892][Error] {} [o.a.f.c.f.FileSystem]: Failed to load a file system via services
java.util.ServiceConfigurationError: org.apache.flink.core.fs.FileSystemFactory: Provider org.apache.flink.fs.s3hadoop.S3AFileSystemFactory could not be instantiated
        at java.util.ServiceLoader.fail(Unknown Source) ~[?:?]
        at java.util.ServiceLoader$ProviderImpl.newInstance(Unknown Source) ~[?:?]
        at java.util.ServiceLoader$ProviderImpl.get(Unknown Source) ~[?:?]
        at java.util.ServiceLoader$3.next(Unknown Source) ~[?:?]
        at org.apache.flink.core.plugin.PluginLoader$ContextClassLoaderSettingIterator.next(PluginLoader.java:103) ~[flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.shaded.guava18.com.google.common.collect.Iterators$5.next(Iterators.java:558) ~[flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.shaded.guava18.com.google.common.collect.TransformedIterator.next(TransformedIterator.java:48) ~[flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.core.fs.FileSystem.addAllFactoriesToList(FileSystem.java:1068) [flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.core.fs.FileSystem.loadFileSystemFactories(FileSystem.java:1050) [flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.core.fs.FileSystem.initialize(FileSystem.java:325) [flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.runtime.taskexecutor.TaskManagerRunner.runTaskManagerSecurely(TaskManagerRunner.java:315) [flink-dist_2.11-1.11.1.jar:1.11.1]
        at org.apache.flink.runtime.taskexecutor.TaskManagerRunner.main(TaskManagerRunner.java:297) [flink-dist_2.11-1.11.1.jar:1.11.1]
Caused by: java.lang.NoClassDefFoundError: Could not initialize class org.apache.flink.fs.s3hadoop.S3FileSystemFactory
        at jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method) ~[?:?]
        at jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance(Unknown Source) ~[?:?]
        at jdk.internal.reflect.DelegatingConstructorAccessorImpl.newInstance(Unknown Source) ~[?:?]
        at java.lang.reflect.Constructor.newInstance(Unknown Source) ~[?:?]
        ... 11 more

【问题讨论】:

    标签: apache-kafka apache-flink


    【解决方案1】:

    正如您所说,Flink 插件通过其自己的类加载器加载,并且与任何其他插件完全隔离。

    如果我们深入研究源代码,则在集群启动时使用另一个密钥(不幸的是,它没有在任何地方记录):

    plugin.classloader.parent-first-patterns.additional
    

    这让您可以使用 PluginClassLoader 将外部 jar 添加到类路径中

    声明+用法:https://github.com/apache/flink/blob/53a4b4407816c2780fed2f8995affbebc1f58c3c/flink-core/src/main/java/org/apache/flink/configuration/CoreOptions.java#L156-L174

    将以下内容添加到 flink-conf.yaml。

    plugin.classloader.parent-first-patterns.additional: org.apache.kafka
    

    这应该可以解决问题

    【讨论】:

    • 问题似乎是说这些属性已经尝试过
    • 这是问题的属性: classloader.parent-first-patterns.additional ,我建议使用另一个属性。 plugin.classloader.parent-first-patterns.additional
    • 使用带有“plugin”前缀的属性确实有效。
    猜你喜欢
    • 1970-01-01
    • 1970-01-01
    • 1970-01-01
    • 1970-01-01
    • 1970-01-01
    • 1970-01-01
    • 1970-01-01
    • 2020-05-20
    • 2018-11-04
    相关资源
    最近更新 更多