Overview

s3cmd is a command-line tool for S3-compatible object storage. It works with Ceph RGW through custom endpoint configuration, making it a popular choice for scripting and admin tasks on Ceph clusters.

Install s3cmd

# On Debian/Ubuntu
apt-get install s3cmd

# On RHEL/CentOS (from the EPEL RPM repository)
dnf install s3cmd

# Via pip
pip install s3cmd

Configure s3cmd for Ceph RGW

Run the interactive configuration:

s3cmd --configure

Or create the config file directly at ~/.s3cfg. For the common path-style setup against a Rook RGW service, point both host_base and host_bucket at the RGW endpoint:

[default]
access_key = myaccesskey
secret_key = mysecretkey
host_base = rook-ceph-rgw-my-store.rook-ceph:80
host_bucket = rook-ceph-rgw-my-store.rook-ceph:80
use_https = False
signature_v2 = False

Basic Operations

Create a bucket:

s3cmd mb s3://my-bucket

Upload a file:

s3cmd put /tmp/report.pdf s3://my-bucket/reports/report.pdf

List objects:

s3cmd ls s3://my-bucket/

Download a file:

s3cmd get s3://my-bucket/reports/report.pdf /tmp/report-downloaded.pdf

Delete an object:

s3cmd del s3://my-bucket/reports/report.pdf

Sync Local Directory to Ceph

s3cmd sync /var/data/ s3://my-bucket/data/ \
  --delete-removed \
  --progress

Set Object ACL

Make an object publicly readable:

s3cmd setacl s3://my-bucket/public-file.txt --acl-public

Restrict to owner only:

s3cmd setacl s3://my-bucket/private-file.txt --acl-private

Get Bucket and Object Info

s3cmd info s3://my-bucket
s3cmd info s3://my-bucket/reports/report.pdf

Multipart Upload for Large Files

s3cmd put /var/data/large-backup.tar.gz s3://my-bucket/backups/ \
  --multipart-chunk-size-mb=100

Disk Usage

s3cmd du s3://my-bucket
s3cmd du s3://my-bucket/reports/

Summary

s3cmd is a versatile CLI tool that works with Ceph RGW by pointing host_base and host_bucket at the RGW endpoint. It supports common S3 operations and is well-suited for shell scripts that need to interact with Ceph object storage.