Even though I have been running Vcenter and ESX for years, I am only recently implementing HA. A few months ago, I enabled it on my smaller cluster and just accepted the defauts for Admission control. This is a 2 node cluster, so it was set to allow 1 node to fail. I didn't think anything else of it as it just worked.
However, now I am trying to enable it on my bigger cluster, and the same settings do not work. It says I do not have resources for the same 1 node failure default settings. In looking at the "Advanced runtime info" it said I only had 62 slots of which only 8 were free. Therefore it could not tolerate even the loss of one node. Now this cluster has 48 processors and 256GB of RAM and is only running at about 25% CPU and 33% RAM on average, so I found this to be strange. Furthermore, my smaller cluster above with 16 processors and 64GB of RAM has 144 slots of which 120 are available, despite having higher overall utilization. Seems odd to me that VMware would calculate less slots on a cluster with literally 4-5x the resources.
So what I have done for the larger cluster is switch to the admission control policy that is percent based, and chose a percentage that leaves me just slightly overutilized if I lost one node. That seems like it should be fine, but I do wonder if I need to fix something in regards to the default not working. Is there any downside to using the percentage based?
