ESXi Hosts Intermittently Show "Not Responding" in vCenter Despite Remaining Pingable and VMs Staying Operational

Products

• VMware vCenter Server

Issue/Introduction

ESXi hosts periodically disconnect from vCenter Server and show as "Not Responding" while:

• Hosts remain pingable from vCenter
• Virtual machines continue running
• Host management interface becomes inaccessible
• vSphere Client shows connectivity errors
• Storage plugin loading fails with "URL unreachable" errors

Environment

• VMware vCenter Server
• VMware ESXi
• Hosts and vCenter in same subnet
• Multiple affected hosts in cluster

Cause

Network path interruption between vCenter Server and ESXi hosts, typically caused by:

• Firewall rules blocking management traffic
• Network segmentation
• Port blocking
• MTU mismatches
• Network hardware issues

Resolution

1. Verify basic connectivity:
   1. Ping ESXi hosts from vCenter using both FQDN and IP
   2. Test DNS resolution in both directions
   3. Run `vmkping` tests from hosts to vCenter

      Test basic connectivity
      vmkping -I vmk0 vcenter-ip

      # Test with specific interface and regular MTU
      vmkping -I vmk0 vcenter-ip -d -s 1472
      
      # For hosts with jumbo frames configured
      # Start with standard size then increase gradually
      vmkping -I vmk0 vcenter-ip -d -s 1472
      vmkping -I vmk0 vcenter-ip -d -s 4972
      vmkping -I vmk0 vcenter-ip -d -s 8972
      
      # Testing other management interfaces if present
      vmkping -I vmk1 vcenter-ip -d -s 1472
      vmkping -I vmk2 vcenter-ip -d -s 1472
      
      # Testing with different packet counts to check for intermittent drops
      vmkping -I vmk0 vcenter-ip -c 100

   4. Verify hosts can reach default gateway
      
      
2. Check management traffic:

   tcpdump src <host-ip> and port 902 -nn

      Verify heartbeat packets are flowing between host and vCenter
   
   
3. Test required ports are open:
   • TCP/UDP 902: Agent communication
   • TCP 443: Secure communication
   • TCP 8043: Host management
   • TCP 8080: Web services
   • TCP 5988/5989: CIM server
     
     
4. Monitor vpxd.log for network-related errors:

   journalctl -b | grep -i "HostSyncFailed"

5. If network path issues identified:
   1. Work with network team to verify firewall rules
   2. Check for network device issues
   3. Verify MTU consistency across path
   4. Test bi-directional connectivity
   5. Remove any blocking rules
      
      
6. If network verification complete but issues persist:
   1. Review host services status
   2. Check vpxa service logs
   3. Verify certificates
   4. Test alternate network paths

Additional Information

• Troubleshooting vCenter Server
• Troubleshooting Networking
• VMware Ports and Protocols