Kafka连接JDBC源OutOfMemoryError

Question

我正在使用Confluent JDBC-Source connector运行以下作业

{
    "name": "jobName",
    "config": {
        "connector.class": "io.confluent.connect.jdbc.JdbcSourceConnector",
        "mode": "timestamp",
        "timestamp.column.name": "dateColumn",
        "topic.prefix": "connect-test.",
        "connection.password": "pw",
        "tasks.max": "1",
        "connection.user": "un",
        "poll.interval.ms": "300000",
        "name": "jobName",
        "connection.url": "jdbc:sqlserver://serverName;Database=dbName;user=un;password=pw",
        "table.whitelist": "tableName"
    }
}

我有一个类似的Kafka-Connect作业在相同的数据库和相同的用户上成功运行，但使用另一个较小的表。所以连接不是问题所在。

在运行作业的Kafka-connect服务器上的日志中，我看到了以下内容：

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "KafkaBasedLog Work Thread - connect-configs"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "kafka-coordinator-heartbeat-thread | connect-cluster"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "kafka-producer-network-thread | producer-1"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "KafkaBasedLog Work Thread - connect-offsets"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "kafka-producer-network-thread | producer-3"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "pool-4-thread-1"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "kafka-producer-network-thread | producer-2"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "KafkaBasedLog Work Thread - connect-status"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "DistributedHerder"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "pool-1-thread-5"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "org.eclipse.jetty.server.session.HashSessionManager@2c243a24Timer"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "qtp1434297727-21"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "qtp1434297727-26"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "qtp1434297727-27"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "kafka-producer-network-thread | producer-14"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "kafka-producer-network-thread | producer-15"

Exception: java.lang.OutOfMemoryError thrown from the UncaughtExceptionHandler in thread "pool-1-thread-4"

所以，没有太多可以继续下去的。运行此作业的服务器现在没有响应，并且不响应REST调用。有什么想法吗？

Answer 1

你需要

1. 通过SSH登录到每个连接工作进程，或者无论您如何到达那里，
2. 会强制停止Kafka连接进程，使其大于运行Kafka连接的每个服务器上的默认值。例如KAFKA_HEAP_OPTS="-Xms512M -Xmx4G"。假设使用Linux，那么对于kafka用户来说，最好的地方应该是bashrc文件。
3. 以您为

导出变量的用户身份再次运行connect-distributed

您可能还想导出更多的值，以便能够公开JMX。通过这种方式，您可以在进程崩溃之前对其进行监视