feat(spot_termination): get termination info from aws not imds

Author: mmcgarr

config/plugin/spot_termination.sh

@@ -4,18 +4,34 @@ OK=0
 NONOK=1
 UNKNOWN=2
 
-TOKEN=$(curl -X PUT "http://169.254.169.254/latest/api/token" \
-  -H "X-aws-ec2-metadata-token-ttl-seconds: 21600" \
-  --max-time 3 --silent --fail 2>/dev/null)
+# Get instance ID from Kubernetes node providerID
+if [ -z "${NODE_NAME}" ]; then
+    exit $UNKNOWN
+fi
 
-status_code=$(curl --max-time 3 --silent --output /dev/stderr --write-out "%{http_code}" \
-    -H "X-aws-ec2-metadata-token: $TOKEN" \
-    "http://169.254.169.254/latest/meta-data/spot/instance-action")
+provider_id="$(curl -s -H "Authorization: Bearer $(cat /var/run/secrets/kubernetes.io/serviceaccount/token)" \
+  --cacert /var/run/secrets/kubernetes.io/serviceaccount/ca.crt \
+  "https://${KUBERNETES_SERVICE_HOST}:${KUBERNETES_SERVICE_PORT}/api/v1/nodes/${NODE_NAME}" \
+  2>/dev/null | jq -r '.spec.providerID')"
 
-if [ "${status_code}" -eq "404" ]; then
+if [ -z "${provider_id}" ]; then
+    exit $UNKNOWN
+fi
+
+# Extract instance ID from providerID (format: aws:///region/instance-id)
+instance_id="$(echo "${provider_id}" | awk -F'/' '{print $NF}')"
+
+if [ -z "${instance_id}" ]; then
+    exit $UNKNOWN
+fi
+
+# Check for spot instance interruption via EC2 API
+interruption_time=$(aws ec2 describe-instances --instance-ids "${instance_id}" \
+  --query 'Reservations[0].Instances[0].DisruptionTime' \
+  --output text 2>/dev/null)
+
+if [ "${interruption_time}" = "None" ] || [ -z "${interruption_time}" ]; then
     exit $OK
-elif [ "${status_code}" -eq "200" ]; then
-    exit $NONOK
 else
-    exit $UNKNOWN
+    exit $NONOK
 fi