Fix bug during cluster health check

[kkworden]

Uses existing code to check if the node should be excluded during the desired node count calculation

[js-timbirkett]

@crhuber - any chance the build/release for this can be kicked. Seems the Github action failed and logs are no longer visible.

[js-timbirkett]

I built a container from master and tried it out. This feature is just plain broken. When trying to rolling update with labels set to exclude a couple of nodes you get:

15:46:56  2022-02-09 15:46:56,187 INFO     All k8s nodes are healthy
15:46:56  2022-02-09 15:46:56,188 INFO     Cluster validation passed. Proceeding with node draining and termination...
15:46:56  2022-02-09 15:46:56,188 INFO     Proceeding with node draining and termination...
15:46:56  2022-02-09 15:46:56,189 INFO     Getting k8s nodes...
15:46:56  2022-02-09 15:46:56,378 INFO     Current k8s node count is 45
15:46:56  2022-02-09 15:46:56,378 INFO     Searching for k8s node name by instance id...
15:46:56  2022-02-09 15:46:56,378 INFO     InstanceId i-0b0abed453f717c04 is node ip-100-64-2-147.eu-west-1.compute.internal in kubernetes land
15:46:56  2022-02-09 15:46:56,379 INFO     Cordoning k8s node ip-100-64-2-147.eu-west-1.compute.internal...
15:46:56  2022-02-09 15:46:56,403 INFO     Node cordoned
15:46:56  2022-02-09 15:46:56,403 INFO     Searching for k8s node name by instance id...
15:46:56  2022-02-09 15:46:56,404 INFO     Could not find a k8s node name for that instance id. Exiting
15:46:56  2022-02-09 15:46:56,404 ERROR    Encountered an error when adding taint/cordoning node 
15:46:56  2022-02-09 15:46:56,404 ERROR    Could not find a k8s node name for that instance id. Exiting

As it appears to be trying to find the excluded node.... 🙈

[dgem]

I'd agree with the above. The problem is that @ https://github.com/hellofresh/eks-rolling-update/blob/master/eksrollup/cli.py#L167 the list of nodes has been filtered... so when in runmode 4 (my experience) @ https://github.com/hellofresh/eks-rolling-update/blob/master/eksrollup/cli.py#L200 it tried to find the instance from the filtered node list :(

get_k8s_nodes() could be changed to return both the included and excluded nodes, and perhaps add a guard in before to check if the node is one that should be considered (isn't in the excluded nodes list)

Any thoughts ?