Authentication and ACLs
Requests to the RADOS Gateway (RGW) can be either authenticated or unauthenticated. RGW assumes unauthenticated requests are sent by an anonymous user. RGW supports canned ACLs.
Authentication
Requests are authenticated with AWS Signatures which are derived from the user’s credentials (S3 access key and secret key).
Most S3 clients and AWS SDKs will generate these signatures for you, given the necessary credentials. When issuing raw HTTP requests, these signatures must be added manually.
AWS Signature v4
Please refer to the official documentation in Authenticating Requests (AWS Signature Version 4).
The following values of the x-amz-content-sha256 request header are supported:
Actual payload checksum value
UNSIGNED-PAYLOAD
STREAMING-UNSIGNED-PAYLOAD-TRAILER
STREAMING-AWS4-HMAC-SHA256-PAYLOAD
STREAMING-AWS4-HMAC-SHA256-PAYLOAD-TRAILER
AWS Signature v2
Please refer to the official documentation in Authenticating Requests (AWS Signature Version 2).
Note
While v2 signatures have been deprecated in AWS, RGW continues to support them.
Authentication against OpenStack Keystone
In a radosgw instance that is configured with authentication against OpenStack Keystone, it is possible to use Keystone as an authoritative source for S3 API authentication. To do so, you must set:
the
rgw keystone
configuration options explained in Integrating with OpenStack Keystone,rgw s3 auth use keystone = true
.
In addition, a user wishing to use the S3 API must obtain an AWS-style
access key and secret key. They can do so with the openstack ec2
credentials create
command:
$ openstack --os-interface public ec2 credentials create
+------------+---------------------------------------------------------------------------------------------------------------------------------------------+
| Field | Value |
+------------+---------------------------------------------------------------------------------------------------------------------------------------------+
| access | c921676aaabbccdeadbeef7e8b0eeb2c |
| links | {u'self': u'https://auth.example.com:5000/v3/users/7ecbebaffeabbddeadbeefa23267ccbb24/credentials/OS-EC2/c921676aaabbccdeadbeef7e8b0eeb2c'} |
| project_id | 5ed51981aab4679851adeadbeef6ebf7 |
| secret | ******************************** |
| trust_id | None |
| user_id | 7ecbebaffeabbddeadbeefa23267cc24 |
+------------+---------------------------------------------------------------------------------------------------------------------------------------------+
The thus-generated access and secret key can then be used for S3 API access to radosgw.
Note
Consider that most production radosgw deployments authenticating against OpenStack Keystone are also set up for RGW Multi-tenancy, for which special considerations apply with respect to S3 signed URLs and public read ACLs.
Access Control Lists (ACLs)
RGW supports S3-compatible ACL functionality. An ACL is a list of access grants that specify which operations a user can perform on a bucket or on an object. Each grant has a different meaning when applied to a bucket versus applied to an object:
Permission |
Bucket |
Object |
---|---|---|
|
Grantee can list the objects in the bucket. |
Grantee can read the object. |
|
Grantee can write or delete objects in the bucket. |
N/A |
|
Grantee can read bucket ACL. |
Grantee can read the object ACL. |
|
Grantee can write bucket ACL. |
Grantee can write to the object ACL. |
|
Grantee has full permissions for object in the bucket. |
Grantee can read or write to the object ACL. |
Internally, S3 operations are mapped to ACL permissions thus:
Operation |
Permission |
---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Some mappings, (e.g. s3:CreateBucket
to WRITE
) are not
applicable to S3 operation, but are required to allow Swift and S3 to
access the same resources when things like Swift user ACLs are in
play. This is one of the many reasons that you should use S3 bucket
policies rather than S3 ACLs when possible.