Proactive HA is a new feature Available in vSphere 6.5 released recently. It’s a kind of feature which will even better help you to protect you environment in case of hardware failure.
Currently all of the hardware components are redundant including power supplies, fans, network cards etc. However the most possible cause of whole server failure occurs while one of these theoretically redundant components fails. To better imagine that let’s think about power supply fail. There is still the second one but during there is only one it is much more loaded. (Similar things you can observe with hard disks in a RAID group – the biggest possibility of a disk fail is during RAID re-building).
ProactiveHA will help you protect the environment in such situations. It will detect hardware conditions of a host and allow you to evacuate the VMs before the trivial issue causes the serious outage. For this feature to function, the hardware vendor must participate. Their hardware monitoring solution will advertise the health of the hardware, and vCenter will query that system to get a status of the hardware components such as the fans, memory, and power supplies. vSphere can then be configured to respond according to the failure.
To let it functions there is a new ESXi host state in vSphere 6.5 – Quarantine mode. It’s similar to maintenance mode but it is not as severe as maintenance mode. It’s mean that DRS will attempt to evacuate all VMs from the host, but only if:
- No performance impact on any virtual machine in the cluster
- None of the business rules is disregarded
- Additionally, any soft affinity or-anti-affinity rules will not be overridden by the evacuation. However, DRS will seek to avoid placing any new VMs.
To set the Proactive HA features, find the Partial Failures and Responses section and set how vSphere should respond to partial failures. The options are to place a degraded host into Quarantine Mode, Maintenance Mode, or Mixed Mode.
Mixed mode means that for moderate degradation, the host will be placed into Quarantine Mode. For Severe failures, it will be placed into Maintenance Mode.
For the moment of writing and availibility of vSphere 6.5 GA the supported failure condition types are: