Bug #21128
do not persist aws-chunked content-encoding
0%
Description
When an object is PUT with Content-Encoding: aws-chunked, rgw is persisting this header. The result is that when the same object is GET'ed, it has the Content-Encoding: aws-chunked header, even though the content is not encoded. This incorrect header confuses some clients.
The correct behaviour would be to strip aws-chunked from the CE before this is persisted, which is what AWS does. See their doc http://docs.aws.amazon.com/AmazonS3/latest/API/sigv4-streaming.html
S3 will store the resulting object without the aws-chunked encoding. Therefore, when you retrieve the object it will not be aws-chunked encoded.
And by observation AWS does indeed strip it. See the two attached files which show the headers for rgw vs. AWS.
History
#1 Updated by Dan van der Ster over 6 years ago
Reproducer is here: https://github.com/alexiri/ceph-content-encoding-bug
#2 Updated by Orit Wasserman over 6 years ago
- Assignee set to Orit Wasserman
#3 Updated by Jos Collin over 6 years ago
- Status changed from New to Fix Under Review
PR Created: https://github.com/ceph/ceph/pull/17838
#4 Updated by Casey Bodley over 2 years ago
- Assignee changed from Orit Wasserman to Daniel Gryniewicz
https://github.com/ceph/ceph/pull/17838 was closed by the stale bot. a more specific fix was requested that only avoids persisting this header when it's value is "aws-chunked"