I have setup a kafka zookeeper and 3 brokers on one machine on ec2 with ports 9092..9094 and am trying to consume the topic content from another machine. The ports 2181 (zk), 9092, 9093 and 9094 (servers) are open to the consumer machine. I can even do a bin/kafka-topics.sh --describe --zookeeper 172.X.X.X:2181 --topic remotetopic which gives me
Topic:remotetopic PartitionCount:1 ReplicationFactor:3 Configs:
Topic: remotetopic Partition: 0 Leader: 2 Replicas: 2,0,1 Isr: 2,0,1
Blockquote
However when i do bin/kafka-console-consumer.sh --zookeeper 172.X.X.X:2181 --from-beginning --topic remotetopic I get
WARN Fetching topic metadata with correlation id 0 for topics [Set(remotetopic)] from broker [id:0,host:localhost,port:9092] failed (kafka.client.ClientUtils$)
java.nio.channels.ClosedChannelException
Why is the consumer trying to read from localhost? Is there any option or commandline or default file where this is read from; and I can change it?
Any help would be appreciated!
Had the same problem. Fixed it by setting the property 'advertised.host.name' as instructed by the comments in the kafka configuration (config/server.properties), and restarting the kafka server.
I had same problem , I am using kafka 0.11 and I started producer and broker in my Ubuntu VM and consumer in Windows 10.
I had changed (and uncommented)
advertised.listeners=PLAINTEXT://your.host.name:9092
to
advertised.listeners=PLAINTEXT://<myActual IP address>:9092
example:
advertised.listeners=PLAINTEXT://192.168.150.150:9092
from file config/server.properties at server/broker side (i.e. in my case it was Ubuntu VM)
I have created a python script which allows you to access kafka/zookeeper via ssh tunnel to consume and produce messages from your local machine.
Compatible for Linux & Mac
Usage: kafkatunnel.py [OPTIONS] COMMAND [ARGS]...
Access kafka/zookeeper via ssh tunnel to consume and produce messages from
your local machine
Commands:
aws retrieve kafka/zookeeper ip's from AWS...
manual provide the IP's of your zookeeper/kafka
try it:
https://github.com/simple-machines/kafka-tunnel
I have setup Kafka and Zookeeper in a similar (and even more complex) arrangement and the kafka-console-consumer.sh does work as advertised.
This is very likely a mis-configured Kafka instance (which specifies its zookeeper nodes in the server configuration).
Check the Kafka configuration. By default if the zookeeper instances are NOT specified, it presumes a singular localhost instance. Otherwise the "server.1" ... 2 etc convention must be used to define the zookeeper instances (minimum three being recommended/required to obtain proper quorum).
