One possible reason why the SQL03 VM was not powered back on after a host failure is that the cluster was configured with the default (best effort) VM high availability mode, which does not guarantee the availability of VMs in case of insufficient resources on the remaining hosts. To resolve this issue, I suggest changing the VM high availability mode to guarantee (reserved segments), which reserves some memory on each host for failover of VMs from a failed host. This way, the SQL03 VM will have a higher chance of being restarted on another host in case of a host failure.
To change the VM high availability mode to guarantee (reserved segments), you can follow these steps:
Log in to Prism Central and select the cluster where the SQL VMs are running.
Click on the gear icon on the top right corner and select Cluster Settings.
Under Cluster Services, click on Virtual Machine High Availability.
Select Guarantee (Reserved Segments) from the drop-down menu and click Save.
To configure the environment to ensure any single host failure affects a minimal number of SQL VMs, I suggest using anti-affinity rules, which prevent VMs that belong to the same group from running on the same host. This way, if one host fails, only one SQL VM will be affected and the other SQL VMs will continue running on different hosts.
To create an anti-affinity rule for the SQL VMs, you can follow these steps:
Log in to Prism Central and click on Entities on the left menu.
Select Virtual Machines from the drop-down menu and click on Create Group.
Enter a name for the group, such as SQL Group, and click Next.
Select the SQL VMs (SQL01, SQL02, SQL03) from the list and click Next.
Select Anti-Affinity from the drop-down menu and click Next.
Review the group details and click Finish.
I hope this helps. How else can I help?
https://portal.nutanix.com/page/documents/details?targetId=AHV-Admin-Guide-v6_5:ahv-affinity-policies-c.html
A screenshot of a computer
Description automatically generated with medium confidence
