Configure vSphere HA (High Availability) is not very difficult, but some rules must be followed. If not, you can have a vSphere HA that is not configured correctly.
When you have enough resources for one or two (or more) hosts failovers, but still, you get “Insufficient resources to satisfy vSphere HA failover level on cluster” issue, then you could have a misconfigured vSphere HA.
In this example above, there is 10 ESXi hosts and 1,25Tb (represents more than 50% of usage memory on the cluster) of free memory and only 5% of CPU usage, and still, we get this warning. Should be more than enough resources and ESXi hosts for any host failure.
Checking vSphere HA – Configurations Issues in Monitor Tab we see that we have a warning regarding a wrong HA configuration.
Next, we need to check what is wrong in our configuration.
In the next images, we notice that the “Failure and Response” configurations have a typical setup. Nothing special is here in this section, VMs will power off and restart in one of the available ESXi hosts. Proactive HA is disabled.
Therefore, the issue must be in the Admission Control option.
What is vSphere HA Admission Control?
vSphere HA Admission Control is a feature in vCenter that ensure resources are available in case of a host failover inside a cluster. Admission Control ensure that VMs have resources in case of a host or hosts, failover and VMs have the resources needed in their resource reservations.
The basis for vSphere HA admission control is how many host failures your cluster is allowed to tolerate and still guarantee failover. The host failover capacity can be set in three ways:
- Cluster resource percentage
- Slot policy
- Dedicated failover hosts
HA Host Failures To Tolerate Policy uses slots to perform Admission Control.
- Is base on the slot size – Power on VM + largest CPU + largest memory resources = 1 slot
- Then Hosts resources / Slot size = Smaller slot size that can support.
- After Admission Control check the results with the Admission Control configuration Failover.
– Slot size is calculated.
– HA determinates how many slots each host in the cluster can hold.
– Calculates Current Failover Capacity of the Cluster.
– Compares this to the Configured Failover Capacity.
Check HERE for more details how Slot Policy works and is calculated.
Like we have already seen in the above images, Admission Control has three types control policies.
- Specify Failover Hosts Admission Control Policy
Based on Hosts and pool reservations.
- Percentage of Cluster Resources Reserved Admission Control Policy
Based on CPU and Memory reservations (no resource pools reservations are taking into account for the resources reservations in the Cluster)
- Host Failures Cluster Tolerates Admission Control Policy
Based on VM Power on Slot Policy (taking into account all Power ON VMs and CPU and Memory used or reserved in Pools).
You can read more about vSphere HA Admission Control in VMware knowledge base.
Admission Control in this example Host failures cluster tolerates set with 1, and Host failover capacity was using “Slot Policy (Powered-on VMs)”.
For a Cluster with 10 vSphere hosts, this is more than enough resources for the failover
In this case here the issues were in the above option.
When you select Slot Policy (Powered-on VMs) Admission Control will start resources calculations and VMs slots. If you have significant VMs or big Reservation Pools, the VM slot will also be large , and could not fit in the host failure capacity (if you have 1 or 2 host failover cluster).
So in our cas,e we have some large VMs (between 64 to 128Gb memory) and also some Reservations Pools for those VMs or group of VMs. Slot Policy then calculates that some reservations cannot assure that will be able to provide the reserved resources in case of host failure (not all ESXi hosts have the same CPU and Memory size).
The solution here is to change Admission Control policy to Cluster Resources Percentage policy and for safe side reserve some % for CPU and memory in case of a host failover.
After changing the Admission Control policy HA vSphere is all green without any warnings, and as we can check in the next image, we have 1,25Tb of free memory in the Cluster with a 25% reserved for the Failover Cluster. Also, the Cluster DRS balance is also in a good condition.
For final test, I put two ESXi host of the Cluster in maintenance mode to check if HA vSphere shows any warning about lack of resources, as we can see in the next image no warnings and still with 787Gb free memory. Also the 25% reserved resources for the Failover Cluster.
With this policy change in vSphere HA Admission Control, I fix the problem with the error “Insufficient resources to satisfy vSphere HA failover level on cluster” in the vSphere HA Cluster.
Like is shown above, vSphere HA Admission Control policy can be set in three different ways, and we need to find the best one that fits our Virtual Infrastructure and the size of VMs and amount of resources we have in the Cluster but also the amount of reserved resources we have in the Cluster.
vSphere HA is a feature that is not easy to handle and also is a feature that is not 100% viable from VMware. We all remember the struggle with had with this feature in all vSphere versions. So it is important to know your infrastructure and configure vSphere Ha taking into account the existing resources and also the level of reserved resources used in a Cluster.
Note: Share this article, if you think it is worth sharing.
©2018 ProVirtualzone. All Rights Reserved