> ## Documentation Index > Fetch the complete documentation index at: https://docs.auxia.io/llms.txt > Use this file to discover all available pages before exploring further. # Google Cloud Storage > Connect files in Google Cloud Storage to Auxia with the self-serve setup wizard [Google Cloud Storage](https://cloud.google.com/storage) (GCS) is Google Cloud's object storage service. Auxia's GCS connector copies files from your bucket using **read-only** access granted to a service account unique to your Auxia project, then maps them into Auxia's schema. Auxia never writes to your bucket. Connect in about 10 minutes Fix common errors What's changed ## Features | Feature | Supported | Notes | | - | - | - | | Self-serve setup | ✅ | Wizard in the Auxia Console; generates the exact grant commands | | AI-assisted schema mapping | ✅ | Reads your files and proposes a mapping with a confidence level | | File formats | ✅ | Self-serve: Parquet (recommended), CSV, newline-delimited JSON. With the Auxia team: also Avro, ORC, Delta Lake, Iceberg | | Date-partitioned folders | ✅ | Seven folder date formats, or whole-folder snapshots | | User ID resolution via joins | ✅ | Resolve a table's user ID by joining it to another table | | Incremental sync | Partial | User attribute tables only. Event files sync as daily full refresh | | Data Health monitoring | ✅ | Per-stage status, run history, and row counts | | Capture deletes | ❌ | Files must be append-only | | Row filtering | ✅ | Files are still copied in full; the filter decides what's read into Auxia | | Transfer only | Coming soon | Available for BigQuery today | | Edit a connection after creation | Coming soon | | | Self-serve backfills | Coming soon | Today, request backfills from your Auxia contact | ## Sync overview * **Access:** two read-only roles on the one bucket you connect — **Storage Object Viewer** (`roles/storage.objectViewer`) to read files, and **Storage Legacy Bucket Reader** (`roles/storage.legacyBucketReader`) so the transfer service can look up the bucket's location. * **Cadence:** self-serve connections sync once a day, and a day's data appears the next morning. Connections set up with the Auxia team can check for new files about every hour. [When will my data show up? →](/data-ingestion/concepts/how-ingestion-works#when-will-my-data-show-up) * **Mode:** event files sync as a **full refresh** of each date folder, so re-runs are always safe. User attribute files sync as new rows only. * **Transfer:** files are copied into Auxia-side storage with Google Cloud Storage Transfer Service, then mapped into Auxia's unified tables. ## File requirements The short version — the full rules are in [File source requirements](/data-ingestion/concepts/file-sources): * **One folder per table**, with every file in a folder sharing one schema. * **Date folders** such as `dt=2026-10-02/` let Auxia read only the newest day. * **Append-only:** never overwrite or delete a file; write corrections as new files. * **CSV** needs a header row and comma delimiters. ``` gs://your-bucket/auxia-export/events/dt=2026-10-02/part-0001.parquet ``` ## Limitations * Whole-file compression (`.parquet.gz`, `.csv.gz`) and archives (`.tar.gz`) aren't supported. Client-side encrypted files (such as PGP) are supported only in limited cases. * Avro, ORC, Delta Lake, and Iceberg need the Auxia team rather than the wizard. * More than about 1 TB per run works, but talk to your Auxia contact before committing to a timeline. * Dates in file names (rather than folder names) aren't detected. * A new connection doesn't load history automatically. Contact your Auxia point of contact to load a past date range. * Data in Snowflake or Databricks can be connected by first exporting it to GCS. See [Don't see your source?](/data-ingestion/data-connections#dont-see-your-source) This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.