donglei created KAFKA-13380:
-------------------------------
Summary: When I migrate nodes from cluster A to cluster B, cluster
A reports an error
Key: KAFKA-13380
URL: https://issues.apache.org/jira/browse/KAFKA-13380
Project: Kafka
Issue Type: Bug
Reporter: donglei
We have two kafka clusters：
* ClusterA version：0.10.0.1
* ClusterB version：0.10.2.1
Recently, due to the shortage of resources in cluster B, three machines are
offline from cluster A and ready to be added to cluster B. the offline steps
are as follows:
* 1. Migrate the topic replicas on the three brokers to other brokers.
* 2. Run the `bin/kafka-server-stop.sh` command on the three brokers to close
the Kafka service.
* 3. Clean the installation directory, data directory and log directory of
Kafka service on the three brokers.
When the threee brokers offline from ClusterA, the ClusterA operates normally.
Then, we add the three brokers to cluster B in turn. After adding, the warning
logs in server.log on the three brokers are as follows:
{code:java}
[2021-10-15 11:12:12,140] INFO [Kafka Server 12], started
(kafka.server.KafkaServer)
[2021-10-15 11:12:12,323] WARN Attempting to send response via channel for
which there is no open connection, connection id 5 (kafka.network.Processor)
[2021-10-15 11:12:18,445] WARN Attempting to send response via channel for
which there is no open connection, connection id 1 (kafka.network.Processor)
[2021-10-15 11:12:29,527] WARN Attempting to send response via channel for
which there is no open connection, connection id 3 (kafka.network.Processor)
[2021-10-15 11:12:31,585] WARN Attempting to send response via channel for
which there is no open connection, connection id 1 (kafka.network.Processor)
[2021-10-15 11:12:31,728] WARN Attempting to send response via channel for
which there is no open connection, connection id 10 (kafka.network.Processor)
[2021-10-15 11:12:57,526] WARN Attempting to send response via channel for
which there is no open connection, connection id 0
(kafka.network.Processor){code}
At the same time, we found that all brokers in clusterA also began to report
error logs as follows:
{code:java}
[2021-10-15 11:13:02,187] ERROR Closing socket for xxx:9092-xxxx:49691 because
of error (kafka.network.Processor)
kafka.network.InvalidRequestException: Error getting request for apiKey: 3 and
apiVersion: 2
at
kafka.network.RequestChannel$Request.liftedTree2$1(RequestChannel.scala:95)
at kafka.network.RequestChannel$Request.<init>(RequestChannel.scala:87)
at
kafka.network.Processor$$anonfun$processCompletedReceives$1.apply(SocketServer.scala:488)
at
kafka.network.Processor$$anonfun$processCompletedReceives$1.apply(SocketServer.scala:483)
at scala.collection.Iterator$class.foreach(Iterator.scala:727)
at scala.collection.AbstractIterator.foreach(Iterator.scala:1157)
at scala.collection.IterableLike$class.foreach(IterableLike.scala:72)
at scala.collection.AbstractIterable.foreach(Iterable.scala:54)
at
kafka.network.Processor.processCompletedReceives(SocketServer.scala:483)
at kafka.network.Processor.run(SocketServer.scala:413)
at java.lang.Thread.run(Thread.java:745)
Caused by: java.lang.IllegalArgumentException: Invalid version for API key 3: 2
at
org.apache.kafka.common.protocol.ProtoUtils.schemaFor(ProtoUtils.java:31)
at
org.apache.kafka.common.protocol.ProtoUtils.requestSchema(ProtoUtils.java:44)
at
org.apache.kafka.common.protocol.ProtoUtils.parseRequest(ProtoUtils.java:60)
at
org.apache.kafka.common.requests.MetadataRequest.parse(MetadataRequest.java:96)
at
org.apache.kafka.common.requests.AbstractRequest.getRequest(AbstractRequest.java:48)
at
kafka.network.RequestChannel$Request.liftedTree2$1(RequestChannel.scala:92)
... 10 more
{code}
Is there any error in the way that my broker goes offline from clusterA ? Why
are there no exceptions after the three brokers go offline? Adding them to
clusterB will cause an error in clusterA service?
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
