Avro 向后模式演变抛出 ClassCastException答案

【问题标题】：Avro backward schema evolution throws ClassCastExceptionAvro 向后模式演变抛出 ClassCastException
【发布时间】：2020-12-18 19:01:06
【问题描述】：

我在尝试使用简单的 Java 程序测试 Avro 模式演变时收到了 ClassCastException。

Avro 版本：1.10.0

customer-v1.avsc

{
  "type": "record",
  "namespace": "com.practice.kafka",
  "name": "Customer",
  "doc": "Avro schema for Customer",
  "fields": [
    {"name":  "first_name", "type":  "string", "doc": "Customer first name"},
    {"name":  "last_name", "type":  "string", "doc": "Customer last name"},
    {"name":  "automated_email", "type":  "boolean", "default": true, "doc": "Receive marketing emails or not"}
  ]
}

customer-v2.avsc

{
  "type": "record",
  "namespace": "com.practice.kafka",
  "name": "CustomerV2",
  "doc": "Avro schema for Customer",
  "fields": [
    {"name":  "first_name", "type":  "string", "doc": "Customer first name"},
    {"name":  "last_name", "type":  "string", "doc": "Customer last name"},
    {"name":  "phone_number", "type":  ["null","boolean"], "default": null, "doc": "Optional phone number"},
    {"name":  "email", "type":  "string", "default":  "missing@example.com", "doc":  "Optional email address"}
  ]
}

Program to serialize v1 and deserialize v2

package com.practice.kafka;

import org.apache.avro.file.DataFileReader;
import org.apache.avro.file.DataFileWriter;
import org.apache.avro.io.DatumReader;
import org.apache.avro.io.DatumWriter;
import org.apache.avro.specific.SpecificDatumReader;
import org.apache.avro.specific.SpecificDatumWriter;

import java.io.File;
import java.io.IOException;

public class BackwardSchemaEvolutionSample {

    public static void main(String[] args) {

        // Step 1 - Create specific record
        Customer customer = Customer.newBuilder().setFirstName("John").setLastName("Doe").setAutomatedEmail(false).build();

        // Step 2 - Write specific record to a file
        final DatumWriter<Customer> datumWriter = new SpecificDatumWriter<>();
        try (DataFileWriter<Customer> dataFileWriter = new DataFileWriter<>(datumWriter)) {
            dataFileWriter.create(customer.getSchema(), new File("customer-v1.avro"));
            dataFileWriter.append(customer);
        } catch (IOException e) {
            e.printStackTrace();
        }

        // Step 3 - Read specific record from a file
        final File file = new File("customer-v1.avro");
        final DatumReader<CustomerV2> datumReader = new SpecificDatumReader<>();
        CustomerV2 customerRecord;
        try (DataFileReader<CustomerV2> dataFileReader = new DataFileReader<>(file, datumReader)) {
            customerRecord = dataFileReader.next();
            System.out.println(customerRecord.toString());
        } catch (IOException e) {
            e.printStackTrace();
        }
    }
}

Result

Exception in thread "main" java.lang.ClassCastException: class com.practice.kafka.Customer cannot be cast to class com.practice.kafka.CustomerV2 (com.practice.kafka.Customer and com.practice.kafka.CustomerV2 are in unnamed module of loader 'app')
    at com.practice.kafka.SchemaEvolutionSample.main(SchemaEvolutionSample.java:34)

您能告诉我如何解决此错误吗？

【问题讨论】：

标签： java apache-kafka avro

【解决方案1】：

您定义了 2 种数据类型 Customer 和 Customer2，并且您不能进行任何转换，因为它们没有继承关系。所以 Java 不能进行强制转换，你得到的是ClassCastException。在您的代码中，唯一的解决方案是捕获 ClassCastException 并在 catch 块中将 Customer 转换为 Customer2。

我假设您正在模拟 Kafka 环境中的架构更改。在这种情况下，您将通过添加新字段或删除旧字段来扩展现有的 avro 架构。

只要类的名称保持不变，avro 架构更改将起作用。

【讨论】：