Host shows the vSphere HA status as Network Partitioned due to network problems

Products

• VMware vSphere ESXi

Issue/Introduction

• A VMware vSphere ESXi host reports the vSphere HA status as "Network Partitioned".

• The affected host cannot reach the VMware vCenter Server.

• The affected host cannot reach other ESXi hosts in the cluster.

• /var/log/dfm/fdm.log shows fdm ping to other hosts management IP not reachable.

  YYYY-MM-DD HH:MM:SS.ms Db(167) Fdm[2100862]: [Originator@6876 sub=Cluster opID=clusterManager.cpp:983-###b4d] Beginning ICMP pings every 1000000 microseconds to host-1###8
YYYY-MM-DD HH:MM:SS.ms Wa(164) Fdm[2100870]: [Originator@6876 sub=Election opID=clusterElection.cpp:157-###655c] ReadMsg: [51 times] Received message from bad ip <host-management-ip>(Reason = Not Reachable)- dropping

• Packet captures (pktcap-uw) during ICMP ping tests on management network (vmkping -I <destination ip> -c 10) reveal that requests exit the problematic host's VMNICs, but these identical requests are not received on the VMNICs of destination cluster hosts or the vCenter Server.

• Refer Packet capture on ESXi using the pktcap-uw tool for guidance on how to take packet captures on ESXi.

Environment

• VMware vSphere ESXi
• VMware vCenter Server

Cause

When performing an ICMP ping test, if packet captures reveal that ICMP requests are transmitted from the source host's VMNICs but are not received by the destination host's VMNICs, this suggests that the packets are being dropped within the underlay or physical network infrastructure.

The physical network dropping these communications is leading to vSphere HA showing as Network Partitioned. This is because for vSphere HA to work properly, an ESXi participating in the HA should be able to communicate with other ESXis in the cluster and also with the vCenter. Please see vSphere HA Checklist for reference.

Resolution

Engage physical network team:  If the packet captures confirm that traffic is being dropped within the network path, engage your internal network infrastructure team to investigate the physical layer for the drop traffic.

Additional Information

When a host reports vSphere HA Network Partitioned, but all hosts and the vCenter have confirmed network reachability, then the root cause may be different. You can look into this KB: Host shows the vSphere HA status as Network Partitioned