> For the complete documentation index, see [llms.txt](https://docs.rootcause.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.rootcause.ai/data-sources-and-connectors/connectors/google-cloud-storage.md).

# Google Cloud Storage

Import files from GCS buckets, or query Parquet in them directly.

***

### Prerequisites

* A service account with read access to the bucket (Storage Object Viewer)
* The service account's JSON key file

***

### Setting up the connection

1. In the Import Manager, click the **Google Cloud Storage** card
2. Enter your connection details:

| Field                 | Description                                                                        | Example                                   |
| --------------------- | ---------------------------------------------------------------------------------- | ----------------------------------------- |
| Project ID            | Google Cloud project                                                               | `my-project-123`                          |
| Service Account Email | The service account identity                                                       | `service@project.iam.gserviceaccount.com` |
| Private Key           | Private key (PEM) from the service account JSON, or paste the entire JSON key file |                                           |
| Default Bucket        | Bucket to use for imports (optional)                                               | `my-data-bucket`                          |

3. Click **Test connection** to verify, then **Save & continue**

***

### Importing data

Browse under **Select Bucket, Folder, and File**, or type bucket and object path. Format auto-detects (CSV, Parquet, JSON), sub-folders can be included, and a multi-file import creates a folder source with one child per file.

***

### Import modes

Full Import, Full Import + Live, Recent Rows, Time Period and Query Directly. Rolling windows re-read the path on each sync and need a single object. Query Directly reads Parquet only. See [Connectors](/data-sources-and-connectors/connectors.md).

***

### Troubleshooting

**403 forbidden**

* Grant Storage Object Viewer on the bucket to the service account

**Invalid key**

* Paste the full JSON key file; a truncated PEM is the usual culprit