cloudstack
cloudstack copied to clipboard
Kubernetes nodes with affinity group
ISSUE TYPE
- Improvement Request
COMPONENT NAME
Kubernetes
CLOUDSTACK VERSION
4.16.1
SUMMARY
If I understand correctly, I'm unable to specify affinity group for newly spawned kubernetes nodes. Api (listkubernetescluster) lists affinity group incorrectly (empty array). My feature request is possibility to setup default affinity group for newly spawned VM's or provide workaround.
My use-case is: On our company infrastructure we have multiple groups of machines (with same CPU, RAM etc.) every group is in separate cluster. Each cluster is dedicated for account (affinity group) and I want to run kubernetes on each cluster. But if I scale it up or enable auto-scale, newly spawned VM's aren't member of affinity group, so VM's are spawned on not dedicated cluster.
STEPS TO REPRODUCE
Simply scale kubernetes cluster
EXPECTED RESULTS
Spawn new virtual machine on dedicated cluster defined by affinity group.
ACTUAL RESULTS
New virtual machines is spawned on non-dedicated cluster.
Thanks for opening your first issue here! Be sure to follow the issue template!
cc @Pearl1594 @davidjumani @shwstppr to triage
@tenvecera would having a new parameter to specify affinity group in createKubernetesCluster API and corresponding option in UI make sense for you or looking for something more? cc @Pearl1594 @davidjumani
@shwstppr Hi Abhishek, thanks for your response.
In fact for me is update API sufficient, but with correspodning option in UI this change should help more than only me.
Is it hard to change it? I can imagine that this is more than "only" adding it to API, but also add into scale and so on. In which version can we expect this change? Right now I've set up statically oversized clusters and I'm looking forward to enabling autoscaling.
moved to 4.19.0.0
- create cluster with affinity group (s)
- update cluster with affinity groups (s)
- provision new vm with affinity groups (s)
- list clusters with affinity groups (s) and add affinity group(s) id/name to response
anything else ? @shwstppr @tenvecera
Hello @weizhouapache,
I think that's all. Thank you for continuing to resolve this issue.
We do not have any work in progress for this at the moment so moving it out of 4.19.0.0 milestone