Handle scheduled events immediately in IMDS mode, the same as queue processor mode

[snay2]

Issue #, if available: #361

Description of changes:
Update the IMDS mode handling of scheduled events to start immediately, the same as how queue processor mode currently works:

aws-node-termination-handler/pkg/monitor/sqsevent/scheduled-change-event.go

Line 100 in 814bccc

StartTime: time.Now(),

We could probably remove the date parsing altogether and get rid of EndTime (lines 90-101), since the SQS version does not set it at all and it does not appear to be used. But in the interest of making a small change, I'm leaving that code for now.

Testing:

Inside the NTH repo folder on my local machine, I ran the following commands:

# Create the kind cluster with two worker nodes
kind create cluster --config test/k8s-local-cluster-test/kind-three-node-cluster.yaml

# Install AEMM with a scheduled event starting a few hours in the future
helm template amazon-ec2-metadata-mock eks/amazon-ec2-metadata-mock \
  --set aemm.events.code="instance-reboot" \
  --set aemm.events.notAfter="2022-07-19T23:00:00Z" \
  --set aemm.events.notBefore="2022-07-19T22:25:00Z" \
  --set aemm.events.notBeforeDeadline="2022-07-19T22:30:00Z" \
  --set aemm.events.state="active" \
| kubectl apply -f -

# Verify that AEMM is serving the scheduled event with the correct times per our configuration
kubectl port-forward pod/<AEMM_pod_name> 1338:1338
open http://localhost:1338/latest/meta-data/events/maintenance/scheduled

# Install the latest release of NTH, instruct it to use AEMM, and have it listen for scheduled events but not Spot ITNs
helm template aws-node-termination-handler eks/aws-node-termination-handler \
  --namespace kube-system \
  --set instanceMetadataURL="http://amazon-ec2-metadata-mock-service.default.svc.cluster.local:1338" \
  --set enableSpotInterruptionDraining="false" \
  --set enableScheduledEventDraining="true" \
| kubectl apply -f -

# Verify that NTH received the scheduled event, but that it's in the future
kubectl logs <NTH_pod_name> --namespace kube-system

# Wait a few minutes and verify that no pods have been cordoned or drained
kubectl get nodes

# Build NTH with the changes in this PR
make build-binaries

# Find the repo name and tag of the image just built
docker images

# Import that image into the kind cluster
kind load docker-image bin-build:v1.16.4-10-g8297c93-linux-amd64

# Install the local build of NTH with the same options as before
helm template aws-node-termination-handler ./config/helm/aws-node-termination-handler \
  --namespace kube-system \
  --set image.repository="bin-build" \
  --set image.tag="v1.16.4-10-g8297c93-linux-amd64" \
  --set instanceMetadataURL="http://amazon-ec2-metadata-mock-service.default.svc.cluster.local:1338" \
  --set enableSpotInterruptionDraining="false" \
  --set enableScheduledEventDraining="true" \
| kubectl apply -f -

# Verify that NTH received the scheduled event, and that its StartTime is now
kubectl logs <NTH_pod_name> --namespace kube-system

# Wait a few minutes and verify that the worker nodes get cordoned and drained
# This will show SchedulingDisabled, and k9s will show pods getting restarted
kubectl get nodes

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[snay2]

Oops, forgot to update the unit tests. Stand by...

[pdk27]

LGTM 👍

[pdk27]

🚀