Mismatch in random ports for notebook kernels

When I try to use the notebook kernels in containerized environments I get different ports stored in DSS and in the kubernetes pod is that normal?

The file jupyter-run/jupyter/runtime/kernel-c234fbb1-84dc-44ad-a18e-7ad0aff81702.json shows different ports:

{
  "shell_port": 51875,
  "iopub_port": 38011,
  "stdin_port": 35689,
  "control_port": 41523,
  "hb_port": 57653,
  "ip": "127.0.0.1",
  "key": "f9ffba4e-cd5d99a382680a9e8421cd8a",
  "transport": "tcp",
  "signature_scheme": "hmac-sha256",
  "kernel_name": "py-dku-containerized-venv--k8s-container"
}

from those shown in the pod's logs:

Setup forwarding
Connect with timeout 30000
Use connection file {"shell_port": 35489, "iopub_port": 39419, "stdin_port": 43865, "control_port": 42353, "hb_port": 51247, "ip": "0.0.0.0", "key": "71d9b24a-10aaa34669dd53b810e7c50f", "transport": "tcp", "signature_scheme": "hmac-sha256", "kernel_name": "", "remoteHost": "dataiku-exec-remote-notebook-3wmu7vfe"}

Failed to setup forwarder

In forwarder.log stored in the notebook files inside DSS instance the process of picking random ports is logged and matches the json file above. There is no place in DSS where the ports shown on the kernel pod logs are stored.

How is that possible? Are random ports chosen twice for notebooks?
It works well for visual recipes for example where random chosen ports match correctly

Operating system used: ubuntu 24