monitor runner custom service returns 404 not found for the particular url

[tonystaark]

This returns 404 since the instance metadata url does not have autoscaling option as its path. Therefore will never execute this logic.

Any idea how this was implemented at the first place?

[kayman-mk]

Hm, not sure why you see this. Checked it on my agent and all urls are working as expected. Which one does not work on your instance?

terraform-aws-gitlab-runner/template/gitlab-runner.tftpl

Lines 319 to 403 in 696ebbe

	imds_token=$(curl -sSf -X PUT "http://169.254.169.254/latest/api/token" -H "X-aws-ec2-metadata-token-ttl-seconds: 300")
	instance_id=$(curl -sSf -H "X-aws-ec2-metadata-token: $imds_token" "http://169.254.169.254/latest/meta-data/instance-id")
	region=$(curl -sSf -H "X-aws-ec2-metadata-token: $imds_token" "http://169.254.169.254/latest/meta-data/placement/region")
	# Function to send complete-lifecycle-action
	send_complete_lifecycle_action() {
	echo "Sending complete-lifecycle-action for instance $instance_id..."
	if [ -z "$region" ]; then
	echo "Failed to retrieve AWS region."
	exit 1
	fi
	echo "Current AWS Region: $region"
	# Retrieve the Auto Scaling Group name associated with the instance
	asg_name=$(aws --region $region ec2 describe-tags --filters "Name=resource-id,Values=$instance_id" "Name=key,Values=aws:autoscaling:groupName" --query 'Tags[].Value' --output text)
	if [ -z "$asg_name" ]; then
	echo "Failed to retrieve Auto Scaling Group name."
	exit 1
	fi
	echo "Auto Scaling Group Name: $asg_name"
	# Retrieve the lifecycle hook names associated with the Auto Scaling Group
	lifecycle_hooks=$(aws --region $region autoscaling describe-lifecycle-hooks --auto-scaling-group-name $asg_name --query "LifecycleHooks[?LifecycleTransition=='autoscaling:EC2_INSTANCE_TERMINATING'].LifecycleHookName" --output text)
	if [ -z "$lifecycle_hooks" ]; then
	echo "No lifecycle hooks found for Auto Scaling Group: $asg_name"
	exit 1
	fi
	echo "Lifecycle Hooks: $lifecycle_hooks"
	# This is needed to let CloudWatch Logs catch up.
	echo "Will send complete actions in 10 seconds..."
	sleep 10
	for hook_name in $lifecycle_hooks; do
	echo "Completing lifecycle action for hook: $hook_name"
	aws autoscaling complete-lifecycle-action \
	--lifecycle-hook-name $hook_name \
	--auto-scaling-group-name $asg_name \
	--lifecycle-action-result CONTINUE \
	--instance-id $instance_id \
	--region $region
	if [ $? -eq 0 ]; then
	echo "Lifecycle action completed successfully for hook: $hook_name"
	else
	echo "Failed to complete lifecycle action for hook: $hook_name"
	fi
	done
	}
	# Function to monitor the existence of the Runner process
	monitor_process() {
	systemctl --no-block stop gitlab-runner.service
	sleep 5 # grace period to allow GitLab Runner to shutdown
	while true; do
	status=$(systemctl is-active gitlab-runner.service)
	if [ "$status" = "inactive" ] || [ "$status" = "failed" ]; then
	echo "GitLab Runner service $status. Proceeding with cleanup."
	send_complete_lifecycle_action
	else
	echo "GitLab Runner Service still running, sleeping..."
	fi
	# it usually takes a while to complete the jobs, so don't poll too often
	sleep 30
	done
	}
	# Main function
	main() {
	# Poll whether the instance is in the terminating state using instance metadata API
	imds_token=$(curl -sSf -X PUT "http://169.254.169.254/latest/api/token" -H "X-aws-ec2-metadata-token-ttl-seconds: 300")
	target_lifecycle_state=$(curl -sSf -H "X-aws-ec2-metadata-token: $imds_token" "http://169.254.169.254/latest/meta-data/autoscaling/target-lifecycle-state")
	if [ "$target_lifecycle_state" = "Terminated" ]; then
	# Monitor the existence of the Runner process
	monitor_process
	else
	if [ "${user_data_trace_log}" = "true" ]; then
	echo "Instance target lifecycle state is $target_lifecycle_state. No action required."
	fi
	fi
	}