ceph rgw: bucket和bucket instance

了解rgw内部实现的时候发现，bucket有两种metadata，一个叫bucket，一个叫bucket instance
通过下面命令可以查看到metadata的类型

# radosgw-admin metadata list
[
    "bucket",
    "bucket.instance",
    "otp",
    "user"
]

实际bucket是以bucket名称命名的metadata，而bucket instance是以bucket内部id命名的metadata。

# radosgw-admin metadata list bucket
[
    "abc",
    "test1",
    "u2dir1",
    "image",
    "test",
    "test2",
    "u2dir2",
    "file",
    "123"
]

# radosgw-admin metadata list bucket.instance
[
    "123:92a8697d-a145-44de-acd3-10f4147b07e2.44697.2",
    "test2:92a8697d-a145-44de-acd3-10f4147b07e2.44697.3",
    "image:92a8697d-a145-44de-acd3-10f4147b07e2.44697.4",
    "abc:92a8697d-a145-44de-acd3-10f4147b07e2.44785.1",
    "file:92a8697d-a145-44de-acd3-10f4147b07e2.44697.5",
    "test1:92a8697d-a145-44de-acd3-10f4147b07e2.44697.1",
    "test:92a8697d-a145-44de-acd3-10f4147b07e2.44294.1",
    "u2dir2:92a8697d-a145-44de-acd3-10f4147b07e2.44294.3",
    "u2dir1:92a8697d-a145-44de-acd3-10f4147b07e2.44294.2"
]

bucket对应内部的类型是RGWBucketEntryPoint，其只保存的bucket概要信息。
bucket instance对应内部的数据类型是RGWBucketInfo，其保存的是bucket的详细信息。

# radosgw-admin metadata get bucket:test
{
    "key": "bucket:test",
    "ver": {
        "tag": "_G_IFZpA7rdMI9sa7cXe1BJk",
        "ver": 1
    },
    "mtime": "2020-08-18 02:36:02.676678Z",
    "data": {
        "bucket": {
            "name": "test",
            "marker": "92a8697d-a145-44de-acd3-10f4147b07e2.44294.1",
            "bucket_id": "92a8697d-a145-44de-acd3-10f4147b07e2.44294.1",
            "tenant": "",
            "explicit_placement": {
                "data_pool": "",
                "data_extra_pool": "",
                "index_pool": ""
            }
        },
        "owner": "user1",
        "creation_time": "2020-08-18 02:36:00.161773Z",
        "linked": "true",
        "has_bucket_info": "false"
    }
}

# radosgw-admin metadata get bucket.instance:test:92a8697d-a145-44de-acd3-10f4147b07e2.44294.1
{
    "key": "bucket.instance:test:92a8697d-a145-44de-acd3-10f4147b07e2.44294.1",
    "ver": {
        "tag": "_CwY-Mon5aKwmDK2pzuuJwnR",
        "ver": 1
    },
    "mtime": "2020-08-18 02:36:02.673609Z",
    "data": {
        "bucket_info": {
            "bucket": {
                "name": "test",
                "marker": "92a8697d-a145-44de-acd3-10f4147b07e2.44294.1",
                "bucket_id": "92a8697d-a145-44de-acd3-10f4147b07e2.44294.1",
                "tenant": "",
                "explicit_placement": {
                    "data_pool": "",
                    "data_extra_pool": "",
                    "index_pool": ""
                }
            },
            "creation_time": "2020-08-18 02:36:00.161773Z",
            "owner": "user1",
            "flags": 0,
            "zonegroup": "a2c80c4d-1687-457a-a454-e121c3038e25",
            "placement_rule": "default-placement",
            "has_instance_obj": "true",
            "quota": {
                "enabled": false,
                "check_on_raw": false,
                "max_size": -1,
                "max_size_kb": 0,
                "max_objects": -1
            },
            "num_shards": 0,
            "bi_shard_hash_type": 0,
            "requester_pays": "false",
            "has_website": "false",
            "swift_versioning": "false",
            "swift_ver_location": "",
            "index_type": 0,
            "mdsearch_config": [],
            "reshard_status": 0,
            "new_bucket_instance_id": ""
        },
        "attrs": [
            {
                "key": "user.rgw.acl",
                "val": "AgKBAAAAAwISAAAABQAAAHVzZXIxBQAAAFVzZXIxBANjAAAAAQEAAAAFAAAAdXNlcjEPAAAAAQAAAAUAAAB1c2VyMQUDNgAAAAICBAAAAAAAAAAFAAAAdXNlcjEAAAAAAAAAAAICBAAAAA8AAAAFAAAAVXNlcjEAAAAAAAAAAAAAAAAAAAAA"
            }
        ]
    }
}

bucket和bucket instance都保存在meta pool的 root 命名空间中

# rados ls -p default.rgw.meta -N root
abc
.bucket.meta.123:92a8697d-a145-44de-acd3-10f4147b07e2.44697.2
test1
.bucket.meta.test2:92a8697d-a145-44de-acd3-10f4147b07e2.44697.3
.bucket.meta.test5:92a8697d-a145-44de-acd3-10f4147b07e2.61333.1
.bucket.meta.image:92a8697d-a145-44de-acd3-10f4147b07e2.44697.4
.bucket.meta.test5-2:92a8697d-a145-44de-acd3-10f4147b07e2.105801.1
.bucket.meta.abc:92a8697d-a145-44de-acd3-10f4147b07e2.44785.1
test5-1
.bucket.meta.file:92a8697d-a145-44de-acd3-10f4147b07e2.44697.5
.bucket.meta.test1:92a8697d-a145-44de-acd3-10f4147b07e2.44697.1
.bucket.meta.test5-1:92a8697d-a145-44de-acd3-10f4147b07e2.62015.1
image
test5
.bucket.meta.test:92a8697d-a145-44de-acd3-10f4147b07e2.44294.1
test
test5-2
test2

总结

bucket以bucket name作为名称，保存的bucket概要信息，其中包含bucket内部ID。
bucket instance以bucket内部id命名，保存bucket的详细信息。
二者都保存在<.rgw.meta> pool的 root命名空间中。
至于为什么这么设计，还没有完全理解。如果bucket名称是唯一，为什么还要有一个内部ID？