前言
如果不指定的话，所有的S3用户默认的pool都是同一个，本文提供一个方法，指定某个用户的默认pool，实现用户之间的差异化。
下面以用户bean_lee为例，为其创建pool－bean，希望该用户的所有用户数据写入pool-bean这个pool。
方法
第一步 设置region
通过如下命令可以获取region信息：
radosgw-admin region get > /tmp/region
我们看到region信息如下：
cat /tmp/region |python -mjson.tool
{
"api_name": "",
"default_placement": "default-placement",
"endpoints": [],
"is_master": "true",
"master_zone": "",
"name": "default",
"placement_targets": [
{
"name": "default-placement",
"tags": []
}
],
"zones": [
{
"endpoints": [],
"log_data": "false",
"log_meta": "false",
"name": "default"
}
]
}
我们可以修改placement_targets信息，改成如下：
cat /tmp/region |python -mjson.tool
{
"api_name": "",
"default_placement": "default-placement",
"endpoints": [],
"is_master": "true",
"master_zone": "",
"name": "default",
"placement_targets": [
{
"name": "default-placement",
"tags": []
},
{
"name": "bean_lee",
"tags": [
"pool-bean"
]
}
],
"zones": [
{
"endpoints": [],
"log_data": "false",
"log_meta": "false",
"name": "default"
}
]
}
然后将修改后的region，设置生效：
radosgw-admin region set < /tmp/region
第二步 设置zone信息
可以通过如下指令获取zone信息：
radosgw-admin zone get > /tmp/zone
可以查看zone相关的信息如下：
{ "domain_root": ".rgw",
"control_pool": ".rgw.control",
"gc_pool": ".rgw.gc",
"log_pool": ".log",
"intent_log_pool": ".intent-log",
"usage_log_pool": ".usage",
"user_keys_pool": ".users",
"user_email_pool": ".users.email",
"user_swift_pool": ".users.swift",
"user_uid_pool": ".users.uid",
"system_key": { "access_key": "",
"secret_key": ""},
"placement_pools": [
{ "key": "default-placement",
"val": { "index_pool": ".rgw.buckets.index",
"data_pool": ".rgw.buckets",
"data_extra_pool": ".rgw.buckets.extra"}}]}
我们修改zone信息中的placement_pools信息，添加了一个key值为bean_lee的placement_pools ,最重要的信息是data_pool指定为我们pool。
{
"control_pool": ".rgw.control",
"domain_root": ".rgw",
"gc_pool": ".rgw.gc",
"intent_log_pool": ".intent-log",
"log_pool": ".log",
"placement_pools": [
{
"key": "default-placement",
"val": {
"data_extra_pool": ".rgw.buckets.extra",
"data_pool": ".rgw.buckets",
"index_pool": ".rgw.buckets.index"
}
},
{
"key": "bean_lee",
"val": {
"data_extra_pool": ".rgw.buckets.extra",
"data_pool": "pool-bean",
"index_pool": ".rgw.buckets.index"
}
}
],
"system_key": {
"access_key": "",
"secret_key": ""
},
"usage_log_pool": ".usage",
"user_email_pool": ".users.email",
"user_keys_pool": ".users",
"user_swift_pool": ".users.swift",
"user_uid_pool": ".users.uid"
}
通过如下命令，修改zone的信息
radosgw-admin zone set < /tmp/zone
第三步 修改user 信息
首先获取用户bean_lee的相关信息：
radosgw-admin metadata get user:bean_lee > /tmp/bean_lee
其信息如下：
{ "key": "user:bean_lee",
"ver": { "tag": "_h1Z0FgVKaKIRvq-4ifElmA0",
"ver": 3},
"mtime": 1478767056,
"data": { "user_id": "bean_lee",
"display_name": "bean_lee",
"email": "bean_lee@123.com",
"suspended": 0,
"max_buckets": 1000,
"auid": 0,
"subusers": [
{ "id": "bean_lee:bean_lee",
"permissions": "full-control"}],
"keys": [
{ "user": "bean_lee",
"access_key": "5EJ6XVLI6V3W8TL9R97D",
"secret_key": "2ppUXqDYI52AMZg+h4Oo+VTPgXbiRCFtgZgifGLI"}],
"swift_keys": [
{ "user": "bean_lee",
"secret_key": "111111"},
{ "user": "bean_lee:bean_lee",
"secret_key": "111111"}],
"caps": [],
"op_mask": "read, write, delete",
"default_placement": "",
"placement_tags": [],
"bucket_quota": { "enabled": false,
"max_size_kb": -1,
"max_objects": -1},
"user_quota": { "enabled": false,
"max_size_kb": -1,
"max_objects": -1},
"temp_url_keys": []}}
将内容修改成如下：其中default_placement一行和placement_tags 一行被修改，修改成和region placement_targets 中新增的name和tag
{ "key": "user:bean_lee",
"ver": { "tag": "_h1Z0FgVKaKIRvq-4ifElmA0",
"ver": 3},
"mtime": 1478767056,
"data": { "user_id": "bean_lee",
"display_name": "bean_lee",
"email": "bean_lee@123.com",
"suspended": 0,
"max_buckets": 1000,
"auid": 0,
"subusers": [
{ "id": "bean_lee:bean_lee",
"permissions": "full-control"}],
"keys": [
{ "user": "bean_lee",
"access_key": "5EJ6XVLI6V3W8TL9R97D",
"secret_key": "2ppUXqDYI52AMZg+h4Oo+VTPgXbiRCFtgZgifGLI"}],
"swift_keys": [
{ "user": "bean_lee",
"secret_key": "111111"},
{ "user": "bean_lee:bean_lee",
"secret_key": "111111"}],
"caps": [],
"op_mask": "read, write, delete",
"default_placement": "bean_lee",
"placement_tags": ["pool-bean"],
"bucket_quota": { "enabled": false,
"max_size_kb": -1,
"max_objects": -1},
"user_quota": { "enabled": false,
"max_size_kb": -1,
"max_objects": -1},
"temp_url_keys": []}}
其中default_placement一行和placement_tags 一行被修改，修改成和region placement_targets 中新增的name和tag
通过如下命令使其生效：
radosgw-admin metadata put user:bean_lee < /tmp/bean_lee
第四步 update regionmap
radosgw-admin regionmap update
第五步 重启radosgw
/etc/init.d/radosgw restart
第五步是否必需，我还不太确定，后面会测试
效果
通过s3cmd 命令创建bucket，同时上传100个小文件到bucket。
s3cmd mb s3://bean_pool2
Bucket 's3://bean_pool2/' created
root@BEAN-1:~# seq 0 99 | xargs -I{} -P 40 s3cmd put /var/log/kern.log s3://bean_pool2/{}
/var/log/kern.log -> s3://bean_pool2/32 [1 of 1]
/var/log/kern.log -> s3://bean_pool2/26 [1 of 1]
498 of 498 100% in 0s 23.58 kB/s done
498 of 498 100% in 0s 43.91 kB/s done
/var/log/kern.log -> s3://bean_pool2/36 [1 of 1]
/var/log/kern.log -> s3://bean_pool2/39 [1 of 1]
/var/log/kern.log -> s3://bean_pool2/38 [1 of 1]
498 of 498 100% in 0s 25.48 kB/s done
498 of 498 100% in 0s 19.27 kB/s done
/var/log/kern.log -> s3://bean_pool2/37 [1 of 1]
498 of 498 100% in 0s 14.03 kB/s done
/var/log/kern.log -> s3://bean_pool2/30 [1 of 1]
通过ceph df输出，可以看到如下一行：
pool-bean 17 49800 0 35981M 100
最后的一行OBJECTS 为100，显示了用户的数据 100个对象（消耗了100个ceph底层对象），确实存放在指定的pool中。