CargoShip

Appearance

Verify & restore it ​

An archive you can't get back isn't a backup. This step proves the round trip: inspect the upload, verify its integrity, and pull a file back.

Throughout, replace the URL with your bucket, prefix, and the upload ID that cargoship upload printed.

Inspect the manifest ​

See what's in the upload without downloading anything:

bash
cargoship info s3://my-bucket/archives/uploads/20260721-a1b2c3

This reads the manifest and prints totals (files, bytes, chunks, shards), compression ratio, and encryption status. Add --verbose for per-shard detail or --json for scripting. To list individual files:

bash
cargoship list s3://my-bucket/archives/uploads/20260721-a1b2c3 --pattern '*.fastq'

Verify integrity ​

bash
cargoship verify s3://my-bucket/archives/uploads/20260721-a1b2c3

verify checks the archive against the checksums recorded in the manifest and exits non-zero if anything is missing or corrupt — so it's safe to use in scripts and CI. Use --quick for a fast metadata-only check, or --deep to re-download the stored objects and re-hash them for data-level integrity.

Restore a file ​

Pull a single file back to a local directory:

bash
cargoship restore s3://my-bucket/archives/uploads/20260721-a1b2c3 ./restored \
  --file path/inside/dataset.txt

You can also restore by pattern, by Git commit, or by DVC stage — see Restoring files. If that file returns byte-for-byte, the full pipeline (compress → shard → upload → manifest → restore) is proven working.

Glacier & Deep Archive

If you uploaded to a cold storage class, files must be thawed before download. restore handles this and can estimate retrieval cost first (--max-restore-cost, --tier, --wait). See Restoring files.

Browse interactively (optional) ​

To explore an archive like a filesystem without extracting it:

bash
cargoship shell s3://my-bucket/archives/uploads/20260721-a1b2c3

Then ls, cd, cat, stat, find, and get inside the archive. See Browsing archives.

Next ​