kitekite2020 opened a new issue #6723: Errors in baremetal cluster configuration URL: https://github.com/apache/pulsar/issues/6723 Cannot get cluster list after initializing cluster metadata. Error message is Reason: javax.ws.rs.ProcessingException: Connection refused: localhost/127.0.0.1:8080 Start bookie. Got error message: [main] ERROR org.apache.bookkeeper.discover.ZKRegistrationManager - BookKeeper metadata doesn't exist in zookeeper. Has the cluster been initialized? Try running bin/bookkeeper shell metaformat ## Steps to reproduce I was deploying a multi-cluster on bare metal following instructions from PULSAR 2.5.0. I configured a cluster with 3 nodes of zookeeper. Each node is hosted in a Ubuntu VM. For each node, $ vi conf/zookeeper.conf #insert ZooKeeper cluster servers server.1=10.0.1.50:2888:3888 server.2=10.0.1.51:2888:3888 server.3=10.0.1.52:2888:3888 For zk1, $ mkdir -p data/zookeeper $ echo 1 > data/zookeeper/myid For zk2, $ mkdir -p data/zookeeper $ echo 2 > data/zookeeper/myid For zk3, $ mkdir -p data/zookeeper $ echo 3 > data/zookeeper/myid Then I can start 3 zookeeper successfully. $ bin/pulsar initialize-cluster-metadata --cluster my-cluster --zookeeper 10.0.1.50:2181 --configuration-store 10.0.1.50:2181 --web-service-url http://10.0.1.60:8080,10.0.1.61:8080,10.0.1.62:8080 --broker-service-url pulsar://10.0.1.60:6650,10.0.1.61:6650,10.0.1.62:6650 The ouput log says the cluster is initialized successfully. $ bin/pulsar-admin clusters list I tried to get the cluster list and got the error message above. I tried a few times on other VMs and there is same errors. Not sure if this is expected behavior. I followed the instructions to continue to configure Bookies on another 3 Ubuntu VMs. I planned to deploy bookie and broker on same VM. $ vi conf/bookkeeper.conf zkServers=10.0.1.50:2181,10.0.1.51:2181,10.0.1.52:2181 advertisedAddress=10.0.1.60 When starting bookie, got error message as below. So i think I might get something wrong in zookeeper. [main] ERROR org.apache.bookkeeper.discover.ZKRegistrationManager - BookKeeper metadata doesn't exist in zookeeper. Has the cluster been initialized? Try running bin/bookkeeper shell metaformat. I followed suggested command,bin/bookkeeper shell metaformat. After that, bookie can start and can do bookiesantiy and simplestest test successfully. But I wonder if I did something wrong because later when I did test with broker failover, I got random errors when sending messages and killing one broker. Edit file conf/broker.conf zookeeperServers=10.0.1.50:2181,10.0.1.51:2181,10.0.1.52:2181 configurationStoreServers=10.0.1.50:2181,10.0.1.51:2181,10.0.1.52:2181 advertisedAddress=10.0.1.60 /*ip of the machine */ clusterName=my-cluster brokerDeduplicationEnabled=true Brokers can run successfully and I can send and receive messages. But when i did availability test, sending/receiving messages and killing one broker, I got passed and failed test cases. In failed cases, producer and subscriber tried to rebuilt connections and always got refused. #### System configuration **Pulsar version**: 2.5.0 Ubuntu 18.04
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
