Failed to add S3 init script in job cluster

[User16752239289]

I use below payload to submit my job that include am init script saved on S3. The instance profile and init script worked on interactive cluster. But when I move to job cluster the init script cannot be configure.

{
    "new_cluster": {
        "spark_version": "8.0.x-scala2.12",
        "spark_conf": {
             "spark.databricks.repl.allowedLanguages": "python,sql"
        },
        "aws_attributes": {
            "instance_profile_arn": "<my instance profile>"
        },
        "init_scripts": [
            {
                "s3": {
                    "destination": "s3://<my bucket>/init.sh"
                }
            }
        ],
        "instance_pool_id": "<pool id>",
        "autoscale": {
            "min_workers": 2,
            "max_workers": 6
        }
    },
    "notebook_task": {
        "notebook_path": "<my notebook>"
    }
}

[User16752239289 · Accepted solution]

It is due to the region is missing. For init script saved in S3, the region field is required. The init script section should be like below :

"init_scripts": [
            {
                "s3": {
                    "destination": "s3://<my bucket>/init.sh",
                    "region": ""
                }
            }
        ]

Note that the regions value can be empty " " or the specific region you want to configure.