Pin VMs to HyperCore nodes #2
Description
We should setup server and/or agent VM node affinity, to achieve HA in case one node in HyperCore is restarted or upgraded.
Mail from Dave:
The other thing might be some attempt at setting preferred node affinity so that master nodes are distributed among cluster nodes and same for workers - that could get complicated but maybe something simple like by default will set preferred nodes for master starting at peerID=0 and increasing and worker nodes starting at highest peerID - going down. Maybe something more generic for "anti-affinity" might be the way to go - for this list of VM's ... attempt to distribute them across nodes where possible. (somewhat related - we also somethimes get requests to offer something like vmware DRS which attempts to "re-balance" VM placement across nodes based on node CPU utilization ... have also thought there might be an ansible playbook "workaround" for that.