Podman pod/containers in Oracle Linux lose connectivity

I am setting up a number of rootless containers (grafana and prometheus) in a pod on Oracle Linux 9. This causes some issues - the first being that ansible insists on placing runtime files in /tmp, which is apparently a known cause of problems. Fixed that by manually setting up and passing XDG_RUNTIME_DIR, but even after fixing that, I run into the same problem. Every morning, the connectivity between the containers in the pod - and from the pod to outside entities - will be down (external clients can still access the containers but the containers themselves cannot communicate with