ESXi Host intermittently transitions to Not Responding state due to UDP 902 Heartbeat packet loss in physical network.
Article ID: 435077
Updated On:
Products
Issue/Introduction
ESXi hosts intermittently show as "Not Responding" in vCenter Server, triggering the HostConnectionStateAlarm. This occurs despite the host remaining manageable via SSH or Direct Console User Interface (DCUI). The vpxd.log on vCenter Server indicates missed heartbeats:
YYYY-MM-DDT HH:MM:SS.###Z info vpxd[#######] [Originator@6876 sub=HostCnx opID=CheckforMissingHeartbeats-#######] [VpxdHostCnx] No heartbeats received from host, time since last heartbeat: 71887ms
Environment
VMware vCenter Server 7.x/8.x
VMware ESXi 7.x/8.x
Cause
The vCenter Server fails to receive UDP heartbeat messages on port 902 from the ESXi hosts within the default 60-second window. Packet captures confirm that while the ESXi host transmits heartbeats every 10 seconds, the packets are dropped by the physical underlay infrastructure (firewalls, routers, or congested links) before reaching the vCenter Server.
Resolution
Perform packet capture on the affected host's management VMKernel interface or corresponding uplink (VMNIC).
Refer Performing Rolling Packet Captures on ESXi
The resulting capture can be analyzed in Wireshark with the help of the "Heartbeat Counter" column:
The capture on the left shows heartbeat packets being sent to vCSA but the Heartbeat Counter column in the second capture shows a jump from 144952 to 144959 indicating missing heartbeats.
Feedback
