重新创建DB容器时，Debezium源任务无法重新连接到postgresql DB。

Question

我们有一个kubernetes集群，Debezium作为一个源任务从Postgresql运行并写入kafka。Debezium，postgres和kafka都是在不同的豆荚中运行。当postgres吊舱被删除，kubernetes重新创建吊舱时，debezium吊舱无法重新连接。来自debezium荚的原木：

    2018-07-17 08:31:38,311 ERROR  ||  WorkerSourceTask{id=inventory-connector-0} Task is being killed and will not recover until manually restarted   [org.apache.kafka.connect.runtime.WorkerTask]
    2018-07-17 08:31:38,311 INFO   ||  [Producer clientId=producer-4] Closing the Kafka producer with timeoutMillis = 30000 ms.   [org.apache.kafka.clients.producer.KafkaProducer]

Debezium继续尝试每隔一段时间刷新未处理的消息，但是给出了以下例外：

    2018-07-17 08:32:38,167 ERROR  ||  WorkerSourceTask{id=inventory-connector-0} Exception thrown while calling task.commit()   [org.apache.kafka.connect.runtime.WorkerSourceTask]
    org.apache.kafka.connect.errors.ConnectException: org.postgresql.util.PSQLException: Database connection failed when writing to copy
    at io.debezium.connector.postgresql.RecordsStreamProducer.commit(RecordsStreamProducer.java:151)
    at io.debezium.connector.postgresql.PostgresConnectorTask.commit(PostgresConnectorTask.java:138)
    at org.apache.kafka.connect.runtime.WorkerSourceTask.commitSourceTask(WorkerSourceTask.java:437)
    at org.apache.kafka.connect.runtime.WorkerSourceTask.commitOffsets(WorkerSourceTask.java:378)
    at org.apache.kafka.connect.runtime.SourceTaskOffsetCommitter.commit(SourceTaskOffsetCommitter.java:108)
    at org.apache.kafka.connect.runtime.SourceTaskOffsetCommitter.access$000(SourceTaskOffsetCommitter.java:45)
    at org.apache.kafka.connect.runtime.SourceTaskOffsetCommitter$1.run(SourceTaskOffsetCommitter.java:82)
    at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
    at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:308)
    at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:180)
    at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:294)
    at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
    at java.lang.Thread.run(Thread.java:748)
    Caused by: org.postgresql.util.PSQLException: Database connection failed when writing to copy
    at org.postgresql.core.v3.QueryExecutorImpl.flushCopy(QueryExecutorImpl.java:942)
    at org.postgresql.core.v3.CopyDualImpl.flushCopy(CopyDualImpl.java:23)
    at org.postgresql.core.v3.replication.V3PGReplicationStream.updateStatusInternal(V3PGReplicationStream.java:176)
    at org.postgresql.core.v3.replication.V3PGReplicationStream.forceUpdateStatus(V3PGReplicationStream.java:99)
    at io.debezium.connector.postgresql.connection.PostgresReplicationConnection$1.doFlushLsn(PostgresReplicationConnection.java:246)
    at io.debezium.connector.postgresql.connection.PostgresReplicationConnection$1.flushLsn(PostgresReplicationConnection.java:239)
    at io.debezium.connector.postgresql.RecordsStreamProducer.commit(RecordsStreamProducer.java:146)
    ... 13 more
    Caused by: java.net.SocketException: Broken pipe (Write failed)
    at java.net.SocketOutputStream.socketWrite0(Native Method)
    at java.net.SocketOutputStream.socketWrite(SocketOutputStream.java:111)
    at java.net.SocketOutputStream.write(SocketOutputStream.java:155)
    at java.io.BufferedOutputStream.flushBuffer(BufferedOutputStream.java:82)
    at java.io.BufferedOutputStream.flush(BufferedOutputStream.java:140)
    at org.postgresql.core.PGStream.flush(PGStream.java:553)
    at org.postgresql.core.v3.QueryExecutorImpl.flushCopy(QueryExecutorImpl.java:939)
    ... 19 more

是否有办法让debezium在可用时重新建立与postgres的连接？还是我漏掉了什么配置？

• Debezium版本0.8
• kubernetes版本1.10.3
• postgres版本9.6

Answer 1

看起来这是一个常见的问题，在debezium和kafka中都有开放的特性请求。

https://issues.jboss.org/browse/DBZ-248

https://issues.apache.org/jira/browse/KAFKA-5352

虽然这些都是开放的，但看起来这是预期的行为。

作为一种解决办法，我在部署中添加了这个活性探测

    livenessProbe:
        exec:
          command:
          - sh
          - -ec
          - ipaddress=$(ip addr | grep 'state UP' -A2 | tail -n1 | awk '{print $2}' | cut -f1  -d'/'); reply=$(curl -s $ipaddress:8083/connectors/inventory-connector/status | grep -o RUNNING | wc -l); if [ $reply -lt 2 ]; then exit 1; fi;
        initialDelaySeconds: 30
        periodSeconds: 5

第一个子句获取容器IP地址：

    ipaddress=$(ip addr | grep 'state UP' -A2 | tail -n1 | awk '{print $2}' | cut -f1 -d'/');

第二个子句在响应json中对“运行”实例进行请求和计数：

    reply=$(curl -s $ipaddress:8083/connectors/inventory-connector/status | grep -o RUNNING | wc -l);

如果“than”出现不足两次，则返回退出代码1。

    if [ $reply -lt 2 ]; then exit 1; fi

它似乎正在进行初始测试--即重新启动postgres会触发debezium容器的重新启动。我想像这样的脚本(虽然可能是“robustified”)可以包含在图像中，以便于探测。