java - 在 apache flink 中反序列化 protobuf 事件中的事件时出现问题

Question

我正在我的 Flink 应用程序中读取来自 kinesis 的事件。事件采用 protobuf 格式。如果我在 flink 应用程序中使用 'com.google.protobuf:protobuf-java:3.7.1' ，我就没有问题。但是，如果我将其更改为 'com.google.protobuf:protobuf-java:3.10.0' 我会收到上述堆栈跟踪异常

java.lang.IncompatibleClassChangeError: class com.google.protobuf.Descriptors$OneofDescriptor has interface com.google.protobuf.Descriptors$GenericDescriptor as super class
        at java.lang.ClassLoader.defineClass1(Native Method)
        at java.lang.ClassLoader.defineClass(ClassLoader.java:763)
        at java.security.SecureClassLoader.defineClass(SecureClassLoader.java:142)
        at java.net.URLClassLoader.defineClass(URLClassLoader.java:468)
        at java.net.URLClassLoader.access$100(URLClassLoader.java:74)
        at java.net.URLClassLoader$1.run(URLClassLoader.java:369)
        at java.net.URLClassLoader$1.run(URLClassLoader.java:363)
        at java.security.AccessController.doPrivileged(Native Method)
        at java.net.URLClassLoader.findClass(URLClassLoader.java:362)
        at java.lang.ClassLoader.loadClass(ClassLoader.java:424)
        at java.lang.ClassLoader.loadClass(ClassLoader.java:357)
        at java.lang.Class.getDeclaredMethods0(Native Method)
        at java.lang.Class.privateGetDeclaredMethods(Class.java:2701)
        at java.lang.Class.privateGetPublicMethods(Class.java:2902)
        at java.lang.Class.privateGetPublicMethods(Class.java:2917)
        at java.lang.Class.getMethods(Class.java:1615)
        at org.apache.flink.api.java.typeutils.TypeExtractor.isValidPojoField(TypeExtractor.java:1786)
        at org.apache.flink.api.java.typeutils.TypeExtractor.analyzePojo(TypeExtractor.java:1856)
        at org.apache.flink.api.java.typeutils.TypeExtractor.privateGetForClass(TypeExtractor.java:1746)
        at org.apache.flink.api.java.typeutils.TypeExtractor.privateGetForClass(TypeExtractor.java:1643)
        at org.apache.flink.api.java.typeutils.TypeExtractor.createTypeInfoWithTypeHierarchy(TypeExtractor.java:921)
        at org.apache.flink.api.java.typeutils.TypeExtractor.privateCreateTypeInfo(TypeExtractor.java:781)
        at org.apache.flink.api.java.typeutils.TypeExtractor.createTypeInfo(TypeExtractor.java:735)
        at org.apache.flink.api.java.typeutils.TypeExtractor.createTypeInfo(TypeExtractor.java:731)
        at org.apache.flink.api.common.typeinfo.TypeInformation.of(TypeInformation.java:211)
        at org.apache.flink.api.java.typeutils.ListTypeInfo.<init>(ListTypeInfo.java:45)
        at com.bagi.streaming.serialization.ProtoSchema.getProducedType(ProtoSchema.java:40)
        at org.apache.flink.streaming.connectors.kinesis.serialization.KinesisDeserializationSchemaWrapper.getProducedType(KinesisDeserializationSchemaWrapper.java:57)
        at org.apache.flink.streaming.connectors.kinesis.FlinkKinesisConsumer.getProducedType(FlinkKinesisConsumer.java:363)
        at org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.addSource(StreamExecutionEnvironment.java:1456)
        at org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.addSource(StreamExecutionEnvironment.java:1414)
        at org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.addSource(StreamExecutionEnvironment.java:1396)
        at com.bagi.streaming.StreamProcessor.getKinesisTrackingStream(StreamProcessor.java:101)
        at com.bagi.streaming.StreamProcessor.getKinesisTrackingStream(StreamProcessor.java:110)
        at com.bagi.streaming.StreamProcessor.consumeKinesis(StreamProcessor.java:117)
        at com.bagi.streaming.StreamProcessor.main(StreamProcessor.java:80)
        at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
        at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
        at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
        at java.lang.reflect.Method.invoke(Method.java:498)
        at org.apache.flink.client.program.PackagedProgram.callMainMethod(PackagedProgram.java:529)
        at org.apache.flink.client.program.PackagedProgram.invokeInteractiveModeForExecution(PackagedProgram.java:421)
        at org.apache.flink.client.program.ClusterClient.run(ClusterClient.java:423)
        at org.apache.flink.client.cli.CliFrontend.executeProgram(CliFrontend.java:813)
        at org.apache.flink.client.cli.CliFrontend.runProgram(CliFrontend.java:287)
        at org.apache.flink.client.cli.CliFrontend.run(CliFrontend.java:213)
        at org.apache.flink.client.cli.CliFrontend.parseParameters(CliFrontend.java:1050)
        at org.apache.flink.client.cli.CliFrontend.lambda$main$11(CliFrontend.java:1126)
        at java.security.AccessController.doPrivileged(Native Method)
        at javax.security.auth.Subject.doAs(Subject.java:422)
        at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1844)
        at org.apache.flink.runtime.security.HadoopSecurityContext.runSecured(HadoopSecurityContext.java:41)
        at org.apache.flink.client.cli.CliFrontend.main(CliFrontend.java:1126)

我正在使用flink@1.8.0和'com.twitter:chill-protobuf:0.9.3'。我正在我的 Mac 上本地构建 flink 应用程序 jar。我已经尝试在 3.10.0 和 3.7.1 上使用 protoc 在 3.10.0 上使用 protobuf-java，以防万一。

这是我的解串器

public class ProtoSchema implements DeserializationSchema<List<Event>> {

    @Override
    public List<Event> deserialize(byte[] message) throws IOException {

        List<Event> events = new LinkedList<>();
        InputStream inputStream = new ByteArrayInputStream(message);

        while (true) {
            Event event = Event.parseDelimitedFrom(inputStream);
            if (event != null) {
                events.add(event);
            } else {
                break;
            }
        }
        return events;
    }

    @Override
    public boolean isEndOfStream(List<Event> nextElement) {
        return false;
    }

    @Override
    public TypeInformation<List<Event>> getProducedType() {
        return new ListTypeInfo<>(Event.class);
    }
}

我正在通过这样做插入

StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

Properties consumerConfig = new Properties();
consumerConfig.put(AWSConfigConstants.AWS_CREDENTIALS_PROVIDER, "AUTO");
consumerConfig.put(AWSConfigConstants.AWS_REGION, region);
consumerConfig.put(ConsumerConfigConstants.SHARD_GETRECORDS_INTERVAL_MILLIS, "300");
consumerConfig.put(ConsumerConfigConstants.SHARD_GETRECORDS_RETRIES, "10");
consumerConfig.put(ConsumerConfigConstants.SHARD_GETRECORDS_MAX, "5000");
consumerConfig.put(ConsumerConfigConstants.STREAM_INITIAL_POSITION, "LATEST");

env.addSource(new FlinkKinesisConsumer<>(name, new ProtoSchema(), consumerConfig)).name("KinesisSource");
env.getConfig().registerTypeWithKryoSerializer(Event.class, ProtobufSerializer.class);

Event.class 使用 protoc@3.10.0 和 protobuf-java@3.10.0 从 protobuf 架构编译而来

Answer 1

正如您在 protobuf-java:3.9.0 的评论中所说，对较低版本 (3.8-) 存在二进制不兼容的更改。

向类class Descriptors.OneofDescriptor添加了父类(super class)Descriptors.GenericDescriptor， 哪个 来自客户端类的超接口(interface)的静态字段可能会隐藏从新父类(super class)继承的字段(具有相同的名称)并导致 IncompleteClassChangeError 异常。 More

因此，如果您的类路径上有 protobuf-java:3.9.0+ 并且还有一些较低版本(3.8-)调用此类，您将收到此错误。 (在我的例子中，它来自 hadoop，它有 2.5 protobuf-java 版本，而我的 fat jar 则有 3.10)

解决方案:

1. 您需要遮盖不兼容的依赖项之一protobuf-java more how to shade depedency with gradle
2. 或者使用 3.8 及更低版本作为临时短视解决方案。