You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Heat based Magnum clusters automatically assign soft-anti-affinity server groups to cluster nodes, which is neat.
I see that there is a PR in CAPO with the goal to also do this -> kubernetes-sigs/cluster-api-provider-openstack#1912
I think it would be a good effort to get this into mcapi, as to try and spread out control plane nodes or give the user to control this behaviour for worker nodes, when the previous PR is merged
The text was updated successfully, but these errors were encountered:
This approach makes absolutly sense. The users expect the control plane to be highly avaiable even in one AZ because they have 3 nodes. So from my point of view the control plane server group should be not soft-anti-affinity but anti-affinity by default. WIth the CAPO patch it should be possible for the user to control a per nodegroup affinity setting with a default to soft-anti-affinity.
Heat based Magnum clusters automatically assign soft-anti-affinity server groups to cluster nodes, which is neat.
I see that there is a PR in CAPO with the goal to also do this -> kubernetes-sigs/cluster-api-provider-openstack#1912
I think it would be a good effort to get this into mcapi, as to try and spread out control plane nodes or give the user to control this behaviour for worker nodes, when the previous PR is merged
The text was updated successfully, but these errors were encountered: