An ESXi 6.x host in the vCenter Server inventory has disconnected due to an All Paths Down (APD) situation. An administrator has corrected the APD issue on the host, but it still remains…

Restarting management agents from the Direct Console User Interface (DCUI) restarts the hostd and vpxa services, which are the components responsible for maintaining the connection between the ESXi host and vCenter Server. An APD event can leave these agents in a failed or stalled state even after the storage path is restored, and restarting them forces a fresh registration and reconnection to vCenter.

The esxcli command esxcli system settings advanced set -d /Scsi/FailVMIOonAPD controls whether VM I/O fails during an APD condition; it is a proactive configuration setting and does not reconnect a disconnected host to vCenter.

Modifying /Disk/ApdTokenRetryCount adjusts how many times ESXi retries I/O tokens during an APD event, which is a preventive tuning parameter and has no effect on reconnecting the host after APD is resolved.

The advanced parameter /Misc/APDHandlingEnable toggles the APD handling feature itself; enabling or disabling it does not restore the management agent connection to vCenter.