Now Testing: Access Log Exports to S3
We're now rolling out Early Access to the "Access Log Exports to S3" feature:
With our automated S3 export, you'll be able to get the access logs for your repositories delivered to you periodically. You pick the frequency and the output format, and we'll make the drop hassle-free. You can then import your logs into your favourite Business Intelligence tools to slice dice and analyse your data at scale.
The exports are available in three flavours (formats), from least to most verbose:
- Apache-style: Logs in the "Apache-style" (i.e. web server) format, with the bare essentials for tracking.
- CSV: Comma-Separated Values, with a header line to indicate the available fields.
- JSON (Single): Detailed records as an array in a single JSON object, with additional header fields.
- JSON (Stream): As above, but newline-delimited records and no header, as described on ndjson.org.
- JSON (Stream+TS): As above, but each line contains a timestamped prefix, for tools like SumoLogic.
You'll be able to authenticate to your S3 logs bucket with least-privilege permissions via either Role Assumption to Pre-Shared Keys; that means we only need to write, but not read, the access logs.
We're configuring the exports to be once-per-day for the testing period, but beyond that, we'll be offering different configurable frequencies, such as once-per-month, once-per-week, once-per-hour, etc.
Sound like something you'd be interested in? Please contact us to get exporting today!
Note: This feature is available to Ultra-tier customers, but we'd be happy to let anyone try before committing. Also, with this release, we've removed the deprecated feature of generating raw access logs via the UI, due to performance reasons. The replacement coming up is a Downloads API in the near future.