"Delay for failure response" setting for "Datastore with APD" in vCenter cluster HA

Products

• VMware vSphere ESXi

Issue/Introduction

The purpose of this article is to explain the HA failover behavior for the "Delay for failure response" setting under "All Paths Down (APD) Failure Response" as follows:

vSphere Availability -> Edit Cluster Settings -> Failures and Responses -> Datastore with APD

Environment

VMware vSphere ESXi 8.0.x

VMware vCenter Server 8.0.x

Resolution

Q: When does vSphere HA failover a protected virtual machine if an All Paths Down (APD) condition occurs for the underlying datastore?

In summary, vSphere HA fails over the virtual machine after 140 seconds plus an additional 3 minutes if the APD condition persists without access to the storage.

• An ESXi host detects an APD condition and starts a timer for the APD Timeout (default: 140 seconds).
• If the device does not recover within this timeout and vSphere HA is configured to respond to APD, the "Delay for failure response" timer begins (default: 3 minutes).
• After the "Delay for failure response" timer expires, if the APD condition still exists, vSphere HA executes the configured response (e.g., restarting affected virtual machines on other hosts).

Additional Information

Best practice for vSphere High Availability (HA) specific to APD and PDL storage events