Rclone

​

Before you start

​

Configure

[tapedrive]
type = s3
provider = Other
access_key_id = TAPE4Q7ZK2M9XH3PD8AW
secret_access_key = mZ9vQ3nL8kR2tW6yB4cX7dF1hJ5pS0aG3uE8iO2q
endpoint = https://s3.example.com
region = solana

​

Everyday commands

# list your buckets
rclone lsd tapedrive:

# list objects
rclone ls tapedrive:my-bucket

# upload a folder, adding new and changed files
rclone copy ./photos tapedrive:my-bucket/photos

# make the bucket match the folder, deleting what is gone locally
rclone sync ./photos tapedrive:my-bucket/photos

# check that the bucket matches the folder
rclone check --size-only ./photos tapedrive:my-bucket/photos

# download a folder
rclone copy tapedrive:my-bucket/photos ./photos

# print an object
rclone cat tapedrive:my-bucket/photos/notes.txt

# delete an object, or everything under a prefix
rclone delete tapedrive:my-bucket/photos/old.jpg
rclone purge tapedrive:my-bucket/photos/2019

​

Good to know

• Do not pass --checksum. ETags here are not MD5 hashes, so rclone silently compares sizes instead and misses a changed file of the same length.
• rclone check compares sizes and warns that hashes could not be checked. That warning is expected.
• rclone re-sends every file on every sync, because the gateway does not store modification times. The gateway ignores unchanged content, so a repeat sync costs nothing on chain.
• rclone mkdir reports success but creates nothing. Buckets are created with tape create.
• rclone lsd shows 1970 as the bucket date, because buckets have no creation date.

​

Next

• Hosting a Static Site for deploying a website.
• Integrations for the errors every S3 client shares.