Using "Suspend to Memory" feature in ESXi 7.0 Update 2

Products

• VMware vSphere ESXi

Issue/Introduction

Suspend to memory is a feature that you can choose to use when you update your ESXi hosts. This allows virtual machines to be suspended in memory on the ESXi host being updated instead of migrating VMs to another ESXi host using vMotion. The Suspend to Memory feature requires Quick Boot to ensure the virtual machine memory state is retained across the reboot. 

This feature helps users:

1. To complete the update much faster end to end.
2. Reduce the VM down time when there is not enough resources to vMotion.

Environment

VMware vSphere ESXi 7.0.0

Resolution

There are prerequisite, configuration, and best practices that you should know when using the "Suspend to Memory" feature.

Note: The suspend to memory feature is restricted to patches (hot patch, express patch etc). For instance, this feature cannot be used to upgrade from 7.0 U2 to 7.0 U3.

Unable to choose use Suspend to Memory or the remediation precheck does not allow users to proceed with Suspend to Memory

Cause: There are certain constraints of using this feature, including server hardware and workloads running in the server. 

Workaround: 

1. Check the server hardware is in the supported list. For more information, see Supported Systems in VMware vSphere 7.0 Update 2 (82558).
2. Run a remediation precheck to identify the issues preventing this feature from working. The "Requirements for Using Suspend to Memory" section of the vSphere Product Documentation outlines the requirements and constraints for suspend to memory. If your hardware is unsupported or another requirement cannot be satisfied, disable Suspend to Memory and complete the remediation without using this feature. If you can, address the requirements, then retry the remediation with Suspend to Memory.

Longer than expected time for the host and workload to come out of maintenance mode

Cause: There are multiple parameters that impacts the time to complete, including memory size, number of VMs, software services running in the host.

Workaround: The vSphere High Availability (HA) feature will restart the VMs on another host if the reboot does not complete in the defined timeout. The Suspend to Memory and vSphere High Availability (HA) section of the vSphere Product Documentation provides more details on this behavior and configuring this timeout.

A host performs a full reboot and VM restarted by vSphere High Availability in rare cases

Cause: There are rare cases where the system cannot complete a Quick Boot and has to resort to a full reboot. Memory contents will not be retained after a full reboot and VMs suspended in memory cannot resume from their suspended state. These VMs will experience a fresh power on.

Workaround:

1. Take a memory snapshot of the VM before performing a remediation with suspend to memory.  If the host resorts to a full reboot then the memory state of the VM at the time of the snapshot can be restored.  If HA is enabled this will happen automatically.  If not please see the “Managing Snapshots” section of the vSphere Product Documentation for how to revert the snapshot.
2. If no memory snapshot was taken prior to the remediation, the VM will have a fresh power on and its memory state at suspend time is not retained.  Ensure the VM is operating as expected in this circumstance.