Amazon S3
For a bucket of objects and GiB, with its largest object at GiB, preflight suggests .
- Current source bucket size exceeds the configured Sync threshold (860.00 GiB > 5 GiB).
- Current source bucket object count exceeds the configured Sync threshold (2,400,000 > 100,000).
Change a number to try your bucket. These are preflight's own sentences with Douklar's default thresholds, 5 GiB and 100,000 objects; each install can change them. Storage classes, encryption and keys are checked on the real bucket.
- What moves
- Missing or changed objects
- Your choices
- Worker relay, server-side copy, S3 Batch Operations or a local-file upload; keep object tags if you want them
How it moves
- Worker relay (default): selects missing or changed keys, downloads with a conditional read and uploads to the destination, with a per-object upload journal.
- Server-side copy: conditional
CopyObjectinside AWS, multipart from 8 MiB, with a destination HEAD check bound to the receipt. - S3 Batch Operations: readiness checks, an explicit Prepare step, four manifest modes, held jobs that wait for Start, completion reports and a cost estimate per job.
- Local-file upload: the browser sends files straight to S3 with short-lived presigned URLs bound to bucket, key, checksum and condition.
- Suggests Sync or Batch from thresholds you can change (5 GiB and 100,000 objects by default).
Limits
- The worker relay stages data on the Douklar host and does not read the uploaded bytes back; server-side copy and Batch keep data inside AWS.
- Archived objects (Glacier, Deep Archive, Intelligent-Tiering archive tiers) are refused with restore guidance.
- ACLs, storage class, checksums, Object Lock, retention and legal holds are not copied.
- SSE-C sources are refused for server-side copy.
- A change to the source bucket policy is handed to its admin as an exact statement.
For ongoing replication of new objects, S3 Replication is the better tool.