Allow short name matching

[dcarrion87]

Using SQS Mode when the event comes in it's checking against the FQDN only:

E.g.

2023/06/28 05:10:57 WRN Unable to list Nodes w/ label, falling back to direct Get lookup of node
2023/06/28 05:10:57 ERR Unable to fetch node labels for node 'ip-10-X-X-X.ap-southeast-2.compute.internal'  error="nodes \"ip-10-X-X-X.ap-southeast-2.compute.internal\" not found"

But the kubernetes.io/hostname assigned by K3s servers is the short name as we're running on Ubuntu. E.g. just ip-10-X-X-X.

Don't see any way to instruct the node termination handler to do a short name match. Any ideas and/or thoughts welcomed.

At this stage I have overriden host to fqdn.

[ssoriche]

We're experiencing the same issue with kubernetes 1.24 running under kOps. The node names are now the instance IDs, and node termination handler is trying to apply the taint by the hostname, not the node name.

When looking through the logs, anywhere else node_name is mentioned, the value is correct, the taint application log line is:

ERR Unable to taint node with taint aws-node-termination-handler/asg-lifecycle-termination:asg-lifecycle-term- error="Unable to fetch kubernetes node from API: nodes "i-123456789abcdef.us-west-2.compute.internal" not found

however the next line is:

INF Pods on node node_name=i-123456789abcdef pod_names=...

has the correct node name

I do see this in the logs though:

INF Requesting instance drain event-id=asg-lifecycle-term-
 instance-id=i-123456789abcdef kind=ASG_LIFECYCLE node-name=i-123456789abcdef.us-west-2.compute.internal provider-id=aws:///us-west-2b/i-123456789abcdef

which indicates the node name is wrong, the instance-id and provider-id are correct.

I've enabled the flag --use-provider-id=true

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. If you want this issue to never become stale, please ask a maintainer to apply the "stalebot-ignore" label.

[github-actions]

This issue was closed because it has become stale with no activity.