containerd runtime integration test failing

[aclevername]

failed 3 times in a row: https://github.com/weaveworks/eksctl/runs/3790538994?check_suite_focus=true

times out when waiting for the nodegroup to report healthy

[0] 2021-10-04 16:25:12 [ℹ]  adding identity "arn:aws:iam::<redacted>:role/eksctl-it-crud-floral-rainbow-163-NodeInstanceRole-1NIOK5ODMQKJI" to auth ConfigMap
[0] 2021-10-04 16:25:12 [▶]  saved auth ConfigMap for "test-containerd"
[0] 2021-10-04 16:25:13 [ℹ]  nodegroup "test-containerd" has 0 node(s)
[0] 2021-10-04 16:25:13 [ℹ]  waiting for at least 2 node(s) to become ready in "test-containerd"
...

inspecting kubectl get nodes shows the nodes as not registered

[Skarlso]

🤦 this test should be expected to fail. How the freck did it not fail so far????!! :D

[aclevername]

@Skarlso didn't you add it 😂 ? https://github.com/weaveworks/eksctl/pull/4051/files

[Skarlso]

Okay, now I know what the problem is actually. The test makes no sense. The containerd setting is only ever applied to unmanaged nodegroups anyways, because the setting doesn't exist on the managed nodegroup path.

However, then this should just pass, right? Because the setting is just ignored. Why did it start failing then? No idea... It times out and that's it. I tried testing it locally, works fine. I'm gonna just remove this test, because it doesn't make any sense. The feature is tested through unit tests.

[Skarlso]

@aclevername Yeah, I did. :DDDD And now, I'm gonna delete it. :D

[Skarlso]

Wait. No. Hang on. What am I saying. This should work. :D

[Skarlso]

I got confused because nodegroups IS unmanaged nodegroups. So this should totally work. But it times out on CI.