Nodes restarting in different Availability Zones

[gwvandesteeg]

In the StorageClass definition you're missing an important setting of WaitForFirstConsumer.
This resolves the issue of topology constrained storage backends (such as AWS EBS).

volumeBindingMode: WaitForFirstConsumer

as per: https://kubernetes.io/docs/concepts/storage/storage-classes/

The main cause of this would be where the instance using the volume is restarted for some reason onto a different availability zone. The above mitigates that and constrains the pod to be more accurately restricted to being in the same zone as the storage (AWS prohibits cross zone EBS bindings).

[xr09]

Hey @gwvandesteeg, thanks a lot for reaching out, I'll take a look at this and get it fixed.