0
votes

I am getting following error in the logs when trying to publish first message to a new topic.

[WARN ] [o.a.kafka.clients.NetworkClient][[Producer clientId=producer-1] Error while fetching metadata with correlation id 766890 : {myTopic-1=INVALID_REPLICATION_FACTOR, myTopic-2=INVALID_REPLICATION_FACTOR}] []

The kafka fails hangs at:

"Hashed wheel timer #1" #521 prio=5 os_prio=0 tid=0x00007f932cd7d000 nid=0x199fa in Object.wait() [0x00007f9322b79000]
   java.lang.Thread.State: TIMED_WAITING (on object monitor)
    at java.lang.Object.wait(Native Method)
    at org.apache.kafka.clients.Metadata.awaitUpdate(Metadata.java:177)
    - locked <0x000000047838b990> (a org.apache.kafka.clients.Metadata)
    at org.apache.kafka.clients.producer.KafkaProducer.waitOnMetadata(KafkaProducer.java:903)
    at org.apache.kafka.clients.producer.KafkaProducer.doSend(KafkaProducer.java:794)
    at org.apache.kafka.clients.producer.KafkaProducer.send(KafkaProducer.java:784)

The kafka jar version we are using is: 0.10.0.1

The server.properties on broker is:

broker.id=3
host.name=<>
port=9092

message.max.bytes=20971520
num.partitions=30
auto.create.topics.enable=true

# Replication configurations
default.replication.factor=2
num.replica.fetchers=2
replica.fetch.max.bytes=20971520


log.dirs=/mnt1/data/kafka/kafka-logs-3
log.retention.hours=48
log.flush.interval.ms=10000
log.flush.interval.messages=20000
log.flush.scheduler.interval.ms=2000
log.cleanup.interval.mins=30

zookeeper.connect=<>
zookeeper.connection.timeout.ms=1000000

# Socket server configuration
num.io.threads=8
num.network.threads=8
socket.request.max.bytes=20971520
socket.receive.buffer.bytes=20971520
socket.send.buffer.bytes=20971520
queued.max.requests=32
fetch.purgatory.purge.interval.requests=100
producer.purgatory.purge.interval.requests=100

inter.broker.protocol.version=0.10.2.0
log.message.format.version=0.10.0
delete.topic.enable=true

Why I am not able to send messages? The topic itself is not being created! The producer properties being used are:

Properties props = new Properties();
        props.put("acks", "0");
        props.put("retries", "0");
        props.put("batch.size", "16384");
        props.put("linger.ms", "100");
        props.put("buffer.memory", "33554432");
        props.put("key.serializer", "org.apache.kafka.common.serialization.IntegerSerializer");
        props.put("value.serializer", "org.apache.kafka.common.serialization.StringSerializer");
        props.put("max.request.size", String.valueOf(ByteUnit.MB.toBytes(10)));
        return props;
2

2 Answers

1
votes

How many brokers do you have in the cluster? The default.replication.factor is 2, which means that you must have at least 2 brokers running.

1
votes

I experienced the same problem and on searching through the Kafka codebase I realized that in some instances, Errors.COORDINATOR_NOT_AVAILABLE is reported as Errors.INVALID_REPLICATION_FACTOR

In my case, the actual underlying problem was that the topic specified in the consumer config did not exist

Ref: https://github.com/apache/kafka/blob/322b10964ce71eac81da9e574be7dec59c0fc93d/core/src/main/scala/kafka/server/KafkaApis.scala#L1106