Disable disk based evictions for Kubernetes 1.19

[hakman]

e2e pre-submits and periodic e2e for 1.19. Logs indicate it is related to the eviction manager that cannot correctly retrieve the disk stats:
https://gcsweb.k8s.io/gcs/kubernetes-jenkins/logs/e2e-kops-aws-k8s-latest/1269848278476787712/

W0608 04:40:31.222453    4996 eviction_manager.go:339] eviction manager: attempting to reclaim ephemeral-storage
I0608 04:40:31.222488    4996 container_gc.go:85] attempting to delete unused containers
I0608 04:40:31.225274    4996 image_gc_manager.go:322] attempting to delete unused images
I0608 04:40:31.231317    4996 eviction_manager.go:350] eviction manager: must evict pod(s) to reclaim ephemeral-storage
I0608 04:40:31.231341    4996 eviction_manager.go:368] eviction manager: pods ranked for eviction: kube-proxy-ip-172-20-42-239.ap-northeast-2.compute.internal_kube-system(98b80b03aaa21bd38ed134d9616a3730)
E0608 04:40:31.231362    4996 eviction_manager.go:569] eviction manager: cannot evict a critical pod kube-proxy-ip-172-20-42-239.ap-northeast-2.compute.internal_kube-system(98b80b03aaa21bd38ed134d9616a3730)
I0608 04:40:31.231370    4996 eviction_manager.go:391] eviction manager: unable to evict any pods from the node
I0608 04:40:40.627941    4996 image_gc_manager.go:305] [imageGCManager]: Disk usage on image filesystem is at 100% which is over the high threshold (85%). Trying to free 3774873 bytes down to the low threshold (80%).
E0608 04:40:40.631282    4996 kubelet.go:1231] Image garbage collection failed multiple times in a row: failed to garbage collect required amount of images. Wanted to free 3774873 bytes, but freed 0 bytes
I0608 04:40:40.631730    4996 kubelet_getters.go:173] "Pod status updated" pod="kube-system/kube-proxy-ip-172-20-42-239.ap-northeast-2.compute.internal" status="Running"

I tried to understand what exactly happens, but I am stuck at finding the root cause. For know, I know that:

• Kubernetes 1.18 works OK
• using containerd 1.3.4 as container runtime works OK
• restarting kubelet.service fixes the issue
• using Docker 19.03.x + Kubernetes 1.19 doesn't work

[olemarkus]

/lgtm

Looks like the cluster passed the validation step \o/

[hakman]

@olemarkus Yes, I think I tracked the issue with failing e2e in 1.19 to this. Once I have it passed I will update the description.

[rifelpet]

Can we open a kops GitHub issue that tracks this? And label it blocks-next so that we don’t forget :)
/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: hakman, rifelpet

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [rifelpet]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment