Kubernetes节点无法加入集群：主机路由不存在问题求助

Hey there! Let's work through this "no route to host" issue when joining your worker node to the Kubernetes master. Even though they're on the same network, small routing or connectivity hiccups can block this—let's break it down step by step.

First: Verify Basic Connectivity #

Before diving into routing, confirm the two nodes can actually reach each other:

• On your worker node, run ping <master-node-ip> (replace with your master's actual IP). If this fails, that confirms the connectivity issue is real.
• Reverse it: run ping <worker-node-ip> from the master to check if the master can reach the worker too.

Reading Your route Output (Simplified) #

When you run sudo route, you'll see a table with key columns to focus on:

Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
default         192.168.1.1     0.0.0.0         UG    0      0        0 eth0
192.168.1.0     0.0.0.0         255.255.255.0   U     0      0        0 eth0

• Default Gateway: The default entry should point to your network's router (e.g., 192.168.1.1). If this is missing or incorrect, your node can't route traffic even within the same subnet in some setups.
• Subnet Route: There should be an entry for your local subnet (like 192.168.1.0 with genmask 255.255.255.0). This tells the node to send traffic for that subnet directly through the local interface (e.g., eth0).

Adding a Missing Route #

If your worker node doesn't have a route to the master's IP/subnet, add it manually:

1. First, identify your worker's network interface (run ip link show—it's usually something like eth0 or ens33).
2. If the master is on the same subnet (e.g., 192.168.1.x), add a route to the entire subnet:

   sudo ip route add 192.168.1.0/24 dev <your-interface-name>
   

   Or if you just need to target the master's specific IP:

   sudo ip route add <master-ip>/32 dev <your-interface-name>
   

3. To make this route persistent across reboots:
   • On Ubuntu/Debian: Edit /etc/netplan/00-installer-config.yaml and add a routes section.
   • On RHEL/CentOS: Edit /etc/sysconfig/network-scripts/ifcfg-<interface> and add a GATEWAY or ROUTE entry.

Check Firewalls & Security Groups #

Don't overlook firewalls—this is a common culprit that mimics routing errors:

• On the master node, ensure the Kubernetes API port (6443) is open:

  sudo ufw allow 6443/tcp  # For Ubuntu/Debian
  sudo firewall-cmd --add-port=6443/tcp --permanent && sudo firewall-cmd --reload  # For RHEL/CentOS
  

• On the worker node, open ports required by Kubernetes: 10250 (TCP), 30000-32767 (TCP), and 8472 (UDP if using Flannel CNI).

Verify Hostname Resolution (If Using Hostnames in Join Command) #

If your kubeadm join command uses the master's hostname instead of IP, make sure the worker can resolve it:

• Run nslookup <master-hostname> on the worker. If it doesn't return the correct IP, add an entry to /etc/hosts on the worker:

  echo "<master-ip> <master-hostname>" | sudo tee -a /etc/hosts
  

Once you've worked through these steps, try running the kubeadm join command again. If you still hit issues, share the full output of sudo route from both nodes and the exact error message from the join attempt—we can dig deeper!

内容的提问来源于stack exchange，提问作者IT_novice