CentOS环境下Storm V1.2.1启动报错：无法从种子主机找到Nimbus领导者

Hey there, let’s tackle that "can't find Nimbus leader from seed hosts" error you’re hitting with your Storm 1.2.1 cluster on CentOS. I’ve worked through this exact issue a handful of times, so let’s break down the fixes step by step:

Before diving into configs, make sure the Nimbus daemon is up and running on your seed host:

• On the Nimbus node, run this command to check for the process:
  ps aux | grep nimbus
  You should see a line with org.apache.storm.daemon.nimbus.Nimbus if it’s running.
• If it’s not running, start it (preferably in the background so it persists after your session ends):
  nohup storm nimbus > nimbus.out 2>&1 &
• If it fails to start, check the Nimbus logs immediately—they’ll tell you why. The default log path is $STORM_HOME/logs/nimbus.log (replace $STORM_HOME with your actual Storm installation directory). Look for errors like port conflicts, missing permissions, or ZooKeeper connection issues.

Since you mentioned you have the storm.yaml handy, verify these critical settings are correct across all nodes (Nimbus, Supervisors, UI):

• nimbus.seeds: This must be set to the IP address or hostname of your Nimbus node(s). For example:

  nimbus.seeds: ["192.168.1.100"]
  # Or if using hostnames:
  nimbus.seeds: ["nimbus-node"]
  

  Make sure every node in the cluster has the same value here—no typos, no missing brackets.
• storm.zookeeper.servers: Storm relies on ZooKeeper to track cluster state, so this needs to point to your running ZooKeeper nodes:

  storm.zookeeper.servers: ["zk-node-1", "zk-node-2"]
  

• storm.local.dir: All nodes need this directory configured, and the Storm user must have read/write access to it. For example:

  storm.local.dir: "/var/lib/storm"
  

  Fix permissions if needed: chown -R storm:storm /var/lib/storm (assuming you’re running Storm as the storm user).
• nimbus.host: While Storm 1.x prefers nimbus.seeds, some components still rely on this setting. Set it to the same IP/hostname as your Nimbus seed:

  nimbus.host: "192.168.1.100"
  

If the configs look good, check if other nodes can reach the Nimbus node:

• Ping the Nimbus IP/hostname from a Supervisor or UI node:
  ping <nimbus-ip-or-hostname>
  If hostnames don’t resolve, add entries to /etc/hosts on all nodes mapping the Nimbus hostname to its IP.
• Test if the Nimbus port (default 6627) is accessible:
  telnet <nimbus-ip> 6627
  Or if telnet isn’t installed: nc -zv <nimbus-ip> 6627
  If this fails, the port is blocked by a firewall.

CentOS’s default firewall and SELinux can block Storm’s network traffic:

• Firewall (firewalld): On the Nimbus node, open the required ports:

  # Allow Nimbus port
  firewall-cmd --add-port=6627/tcp --permanent
  # Allow ZooKeeper port (if ZK is on the same node)
  firewall-cmd --add-port=2181/tcp --permanent
  # Reload firewall to apply changes
  firewall-cmd --reload
  

  Repeat for any other ports your cluster uses (like Supervisor ports 6700-6703, UI port 8080 if needed).
• SELinux: Temporarily disable it to test if it’s causing issues:
  setenforce 0
  If the error goes away, you can either permanently disable SELinux (not recommended for production) or configure SELinux rules to allow Storm’s network traffic.

Storm can’t find Nimbus if ZooKeeper isn’t running properly:

• On each ZooKeeper node, check its status:
  zkServer.sh status
  You should see either "Leader" or "Follower"—if it says "Error contacting service", ZooKeeper isn’t running, so start it with zkServer.sh start.
• Test ZooKeeper connectivity from a Storm node:
  zkCli.sh -server <zk-ip>:2181
  If you can connect, check if the Storm root node exists: ls /storm—you should see entries for Nimbus if it’s registered correctly.

Once you’ve fixed any issues above, restart the cluster in the right order:

1. Stop all running Storm daemons (Nimbus, Supervisors, UI) on their respective nodes.
2. Start ZooKeeper (if it wasn’t running).
3. Start Nimbus and wait 30 seconds for it to fully initialize and register with ZooKeeper.
4. Start Supervisors and the UI service.

That should get your cluster up and running. If you’re still hitting issues, share the relevant snippets from your nimbus.log and storm.yaml, and we can dig deeper.

内容的提问来源于stack exchange，提问作者Sellamani