My workspace in MS Fabric is private endpoint enabled so Starter Pools aren't available so I've created a Custom Pool.

My notebooks take 3-5 minutes to start up from cold and don't honour the 20 min timeout set. What I mean is that if I restart the sae pipeline, for example, that contains a notebook, the notebook activity will again take 3-5 minutes to start up.

I've attached screenshots of the custom pool settings.

This is the json of the Pipeline that I'm testing with. As you can see there's only activity.

{
    "name": "pl_Test",
    "objectId": "30445b45-xxx",
    "properties": {
        "activities": [
            {
                "name": "nb_SparkPoolWarmUp",
                "type": "TridentNotebook",
                "dependsOn": [],
                "policy": {
                    "timeout": "0.00:10:00",
                    "retry": 0,
                    "retryIntervalInSeconds": 30,
                    "secureOutput": false,
                    "secureInput": false
                },
                "typeProperties": {
                    "notebookId": "b6dde9a9-xxx",
                    "workspaceId": "29f8113f-xxxx",
                    "sessionTag": {
                        "value": "@replace(pipeline().DataFactory,'-', '_')",
                        "type": "Expression"
                    }
                }
            }
        ],
        "lastModifiedByObjectId": "9a11117d-xxx",
        "lastPublishTime": "2026-03-27T15:04:05Z"
    }
}

And notebook python code

# Welcome to your new notebook

print("This is just to prebake/warmup spark pools to be utlised by notebooks downstream")

This is a screenshot from the Spark Monitoring

How can I better configure so that my notebooks start quicker, execute quicker and honour the timeout setting?