Ceph » rgw

after replication, the bucket index incorrectly uses the compressed size instead of the actual size

tested on two zones with zlib compression enabled, the compress-encrypted zonegroup feature enabled, and a rgw crypt default encryption key configured

primary:

       "rgw.main": {
            "size": 20971520,
            "size_actual": 20971520,
            "size_utilized": 20424,
            "size_kb": 20480,
            "size_kb_actual": 20480,
            "size_kb_utilized": 20,
            "num_objects": 1
        },

~/ceph/build $ s3cmd -c c1.s3cfg ls s3://testbucket
2023-10-24 18:46     20971520  s3://testbucket/20m.iso

~/ceph/build $ s3cmd -c c1.s3cfg info s3://testbucket/20m.iso
s3://testbucket/20m.iso (object):
   File size: 20971520

secondary:

        "rgw.main": {
            "size": 20424,
            "size_actual": 20480,
            "size_utilized": 20424,
            "size_kb": 20,
            "size_kb_actual": 20,
            "size_kb_utilized": 20,
            "num_objects": 1
        }

~/ceph/build $ s3cmd -c c2.s3cfg ls s3://testbucket
2023-10-24 18:46        20424  s3://testbucket/20m.iso

~/ceph/build $ s3cmd -c c2.s3cfg info s3://testbucket/20m.iso
s3://testbucket/20m.iso (object):
   File size: 20971520

the replicated object itself has the correct size and can be downloaded successfully:

~/ceph/build $ s3cmd -c c2.s3cfg get s3://testbucket/20m.iso 20m.iso.out
download: 's3://testbucket/20m.iso' -> '20m.iso.out'  [1 of 1]
 20971520 of 20971520   100% in    0s    44.11 MB/s  done