GKE node pools feature

[Mark White]

Hi,

Just a quick bit of feedback on the GKE node pools feature - it seems it would be very useful to apply custom labels with the pools so you could have a ‘high-cpu’ label on one pool, and a ‘high-mem’ label on another pool (for example), then target the containers based on the labels. If you added a third pool that was also high cpu you could tag it with the same label, which isn’t possible with the existing 'cloud.google.com/gke-nodepool=' label selector where each pool has a different label.

Thanks,

Mark White

Codec

CTO

(07400) 958883

[Tim Hockin]

If you specify the resources you need on your pods, we'll figure out
where to put them. No need for node labelling in the general case.

> --
> You received this message because you are subscribed to the Google Groups
> "Containers at Google" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to google-contain...@googlegroups.com.
> To post to this group, send email to google-c...@googlegroups.com.
> Visit this group at https://groups.google.com/group/google-containers.
> For more options, visit https://groups.google.com/d/optout.

[Mark White2]

I'm talking about situations where for some reason you want to bind particular pods to particular node pools outside the automatic scheduling.

For example I may have a Kubernetes cluster with 4 CPU nodes and pod X that needs to run across 8 CPUs, so I started an extra node-pool with some 8 CPU nodes. Of course pod X should be defined with an 8 CPU resource request, and that may be fine ... but if Kubernetes schedules another pod on one of the 8 CPU machines first then the pod with an 8 CPU request won't be able to get scheduled. 

(better rescheduling in the future should also address this situation, and of course another option in this situation is to have two completely separate clusters rather than using node pools)

Mark.