Monitoring pods through http results in "Name or service not known"

I’ve deployed checkmk as a pod on an EKS cluster A and I’m using it to monitor both EKS cluster A and B (on which I’ve successfully deployed the checkmk collectors).
All deployed pods have been conveniently auto discovered out of the box.
When I however try to create the following test, it fails with Name or service not known error.

To my understanding this test runs as follows:

/omd/sites/cmk/lib/nagios/plugins/check_http -p 8080 ip_address_of_pod

which succeeds when I manually run with either the pod’s ip or the headless service’s fqdn.

It fails however when executed through checkmk and seems that $_HOSTADDRESS_4$ macro is not properly expanded.

Since the monitored service is part of a statefulset exposed through a headless service what I would expect is for $_HOSTADDRESS_4$ to either be the service-name-0.servicename.namespace or the corresponding ip address.

What am I missing here ?