CEPH-4:ceph RadowGW对象存储功能详解
ceph RadosGW对象存储使用详解
一个完整的ceph集群,可以提供块存储、文件系统和对象存储。
本节主要介绍对象存储RadosGw功能如何灵活的使用,集群背景:
$ ceph -s
cluster:
id: f0a8789e-6d53-44fa-b76d-efa79bbebbcf
health: HEALTH_OK
services:
mon: 1 daemons, quorum a (age 2d)
mgr: a(active, since 2d)
mds: cephfs:1 {0=cephfs-a=up:active} 1 up:standby-replay
osd: 1 osds: 1 up (since 2d), 1 in (since 2d)
rgw: 1 daemon active (my.store.a)
data:
pools: 10 pools, 200 pgs
objects: 1.29k objects, 3.5 GiB
usage: 60 GiB used, 798 GiB / 858 GiB avail
pgs: 200 active+clean
io:
client: 852 B/s rd, 1 op/s rd, 0 op/s wr
什么是对象存储
- 对象存储,又称键值存储,通过其接口指令,例如简单的GET、PUT、DEL等,向存储服务上传下载数据;
- 对象存储中所有数据都被认为是一个对象。所以,任何数据都可以存入对象存储中,如图片、视频、音频等;
- 常见的对象存储厂商有Swift、S3等,ceph就支持Swift API和AWS S3两种标准。
ceph对象存储的构成
Ceph对象存储是通过 RGW组件 来实现,什么是 rgw 呢?
- rgw全称Rados Gateway,是一种服务,使客户端能够利用标准对象存储API来访问ceph对象网关;
- ceph 0.8版本之后使用Civeweb的web服务器来响应api请求,说白了,rgw里边就是一个web服务;
- 客户端使用http/https协议通过RESTful API与rgw通信;
- rgw通过librados与ceph集群通信,利用cephx加密协议与ceph存储通信;
- rgw通过bucket来实现数据存储和多用户的隔离;
- 可以部署多个rgw,实现负载均衡及高可用。
ceph RadosGW中有一个bucket桶的概念,一般项目或者分类会使用bucket来进行隔离,bucket的权限控制,想要操作某个bucket,操作用户必须有对此bucket的对应操作权限,bucket最终的数据其实还是通过PG来落盘到后端的osd存储中的。
RadosGW存储池作用
rgw安装流程请参考之前的部署文档,此处不在赘述,默认端口7480,能够curl通就表示安装成功:
$ curl 10.153.204.13:30080
<?xml version="1.0" encoding="UTF-8"?><ListAllMyBucketsResult xmlns="http://s3.amazonaws.com/doc/2006-03-01/"><Owner><ID>anonymous</ID><DisplayName></DisplayName></Owner><Buckets></Buckets></ListAllMyBucketsResult>
我这里更改了默认端口,改为了30080
rgw安装完毕后,会有一些默认创建的存储池:
$ ceph osd lspools | grep rgw
1 .rgw.root
3 my-store.rgw.control
6 my-store.rgw.meta
7 my-store.rgw.log
8 my-store.rgw.buckets.index
9 my-store.rgw.buckets.non-ec
10 my-store.rgw.buckets.data
这些存储池也是分为两种类型的,一种是元数据存储池,一种是数据存储池:
- .rgw.root : 包含realm(领域信息),比如zone和zonegroup。
- rgw.log:存储日志信息,用户记录各种log信息。
- rgw.control:系统控制池,在有数据更新时,通知其它RGW更新缓存。
- rgw.meta:元数据存储池,通过不同的名称空间分别存储不同的rados对象,这些名称空间包括用户的UID,及其Bucket映射信息的名称空间users.uid,用户的密钥名称空间users.keys,用户的emai名称空间users.email,用户的subuser的名称空间 users.swift,bucket的名称空间root等。
- rgw.buckets.index:存放bucket到object的索引信息。
- rgw.buckets.non-ec:数据的额外信息存储池。
- rgw.buckets.data:存放对象的数据
RadosGW常用操作详解
查看全部zone
$ radosgw-admin zone list
{
"default_info": "a06a6df5-68a4-47f0-9afa-2ac1c09aee58",
"zones": [
"my-store"
]
}
默认为default,我这里更改名字叫my-store
查看zone详情
$ radosgw-admin zone get --rgw-zone=my-store
{
"id": "a06a6df5-68a4-47f0-9afa-2ac1c09aee58",
"name": "my-store",
"domain_root": "my-store.rgw.meta:root",
"control_pool": "my-store.rgw.control",
"gc_pool": "my-store.rgw.log:gc",
"lc_pool": "my-store.rgw.log:lc",
"log_pool": "my-store.rgw.log",
"intent_log_pool": "my-store.rgw.log:intent",
"usage_log_pool": "my-store.rgw.log:usage",
"roles_pool": "my-store.rgw.meta:roles",
"reshard_pool": "my-store.rgw.log:reshard",
"user_keys_pool": "my-store.rgw.meta:users.keys",
"user_email_pool": "my-store.rgw.meta:users.email",
"user_swift_pool": "my-store.rgw.meta:users.swift",
"user_uid_pool": "my-store.rgw.meta:users.uid",
"otp_pool": "my-store.rgw.otp",
"system_key": {
"access_key": "",
"secret_key": ""
},
"placement_pools": [
{
"key": "default-placement",
"val": {
"index_pool": "my-store.rgw.buckets.index",
"storage_classes": {
"STANDARD": {
"data_pool": "my-store.rgw.buckets.data"
}
},
"data_extra_pool": "my-store.rgw.buckets.non-ec",
"index_type": 0
}
}
],
"realm_id": ""
}
radosgw创建新用户认证
$ radosgw-admin user create --uid="vfan" --display-name="my vfan"{
"user_id": "vfan",
"display_name": "my vfan",
"email": "",
"suspended": 0,
"max_buckets": 1000,
"subusers": [],
"keys": [
{
"user": "vfan",
"access_key": "Q6VGP3LYMH99D0A9GUV0",
"secret_key": "NVDfq7CBJgpUnCXKqbgVuKvI3siWNbx0sRltClA4"
}
],
"swift_keys": [],
"caps": [],
"op_mask": "read, write, delete",
"default_placement": "",
"default_storage_class": "",
"placement_tags": [],
"bucket_quota": {
"enabled": false,
"check_on_raw": false,
"max_size": -1,
"max_size_kb": 0,
"max_objects": -1
},
"user_quota": {
"enabled": false,
"check_on_raw": false,
"max_size": -1,
"max_size_kb": 0,
"max_objects": -1
},
"temp_url_keys": [],
"type": "rgw",
"mfa_ids": []
}
新建一个子用户
为了给用户新建一个子用户 (Swift 接口) ,必须为该子用户指定用户的 ID(--uid={username}
),子用户的 ID 以及访问级别:
$ radosgw-admin subuser create --uid=vfan --subuser=vfan:swift --access=full
{
"user_id": "vfan",
"display_name": "my vfan",
"email": "",
"suspended": 0,
"max_buckets": 1000,
"subusers": [
{
"id": "vfan:swift",
"permissions": "full-control"
}
],
"keys": [
{
"user": "vfan",
"access_key": "Q6VGP3LYMH99D0A9GUV0",
"secret_key": "NVDfq7CBJgpUnCXKqbgVuKvI3siWNbx0sRltClA4"
}
],
"swift_keys": [
{
"user": "vfan:swift",
"secret_key": "GrjjD8yJgr2khUCIeRmggNMWqnganFlhMKMMom9s"
}
],
"caps": [],
"op_mask": "read, write, delete",
}
--access=full并不仅仅代表读写,因为他还包括访问权限策略。
查看user列表
$ radosgw-admin user list
[
"vfan",
"ceph-object-user"
]
禁用或启动一个用户
创建账户后,默认是启用状态,可以将其设置为关闭状态:
## 停用一个用户
$ radosgw-admin user suspend --uid=vfan
## 启用一个用户
$ radosgw-admin user enable --uid=vfan
主要是用户中的"suspended"值发生了变化,开启为0,关闭为1。
添加或删除 用户管理权限
## 添加
$ radosgw-admin caps add --uid=vfan --caps="users=*"
## 删除
$ radosgw-admin caps rm --uid=vfan --caps="users=write"
--caps="[users|buckets|metadata|usage|zone]=[*|read|write|read, write]"
删除用户 或 子用户
## 删除用户
$ radosgw-admin user rm --uid=vfan
## 删除子用户
$ radosgw-admin subuser rm --subuser=vfan:swift
查看所有的bucket桶
$ radosgw-admin bucket list
[
"my-test-bucket"
]
查看桶内对象
$ radosgw-admin bucket list --bucket=my-test-bucket
[
{
"name": "hello.txt",
"instance": "",
"ver": {
"pool": 10,
"epoch": 1
},
"locator": "",
"exists": "true",
"meta": {
"category": 1,
"size": 12,
"mtime": "2022-03-30T10:51:38.420295Z",
"etag": "ed076287532e86365e841e92bfc50d8c",
"storage_class": "",
"owner": "vfan",
"owner_display_name": "my vfan",
"content_type": "application/octet-stream",
"accounted_size": 12,
"user_data": "",
"appendable": "false"
},
"tag": "a06a6df5-68a4-47f0-9afa-2ac1c09aee58.24132.17942",
"flags": 0,
"pending_map": [],
"versioned_epoch": 0
}
]
查看存储桶详情
$ radosgw-admin bucket stats --bucket=my-test-bucket
{
"bucket": "my-test-bucket",
"num_shards": 0,
"tenant": "",
"zonegroup": "fd710024-4ba3-41bb-9f96-579d8f03dd1b",
"placement_rule": "default-placement",
"explicit_placement": {
"data_pool": "",
"data_extra_pool": "",
"index_pool": ""
},
"id": "a06a6df5-68a4-47f0-9afa-2ac1c09aee58.24134.1",
"marker": "a06a6df5-68a4-47f0-9afa-2ac1c09aee58.24134.1",
"index_type": "Normal",
"owner": "vfan",
"ver": "0#2",
"master_ver": "0#0",
"mtime": "2022-03-30T10:51:38.323147Z",
"creation_time": "2022-03-30T10:51:38.321498Z",
"max_marker": "0#",
"usage": {
"rgw.main": {
"size": 12,
"size_actual": 4096,
"size_utilized": 12,
"size_kb": 1,
"size_kb_actual": 4,
"size_kb_utilized": 1,
"num_objects": 1
}
},
"bucket_quota": {
"enabled": false,
"check_on_raw": false,
"max_size": -1,
"max_size_kb": 0,
"max_objects": -1
}
}
查看用户配额
$ radosgw-admin user info --uid=vfan | grep -A 5 "quota"
"bucket_quota": {
"enabled": false,
"check_on_raw": false,
"max_size": -1,
"max_size_kb": 0,
"max_objects": -1
--
"user_quota": {
"enabled": false,
"check_on_raw": false,
"max_size": -1,
"max_size_kb": 0,
"max_objects": -1
默认这些配额都是未激活的,处于false状态。
激活用户配额
$ radosgw-admin quota enable --quota-scope=user --uid=vfan
$ radosgw-admin user info --uid=vfan | grep -A 5 "quota"
"bucket_quota": {
"enabled": false,
"check_on_raw": false,
"max_size": -1,
"max_size_kb": 0,
"max_objects": -1
--
"user_quota": {
"enabled": true,
"check_on_raw": false,
"max_size": -1,
"max_size_kb": 0,
"max_objects": -1
已激活用户配额,此时可以修改最大限额,默认是不限制。
更新配额
$ radosgw-admin quota set --uid=vfan --quota-scope=user --max-objects=10000 --max-size=107374182400
[cephadmin@yq01-aip-aikefu10.yq01.baidu.com ~]$ radosgw-admin user info --uid=vfan | grep -A 5 "quota"
"bucket_quota": {
"enabled": false,
"check_on_raw": false,
"max_size": -1,
"max_size_kb": 0,
"max_objects": -1
--
"user_quota": {
"enabled": true,
"check_on_raw": false,
"max_size": 107374182400,
"max_size_kb": 104857600,
"max_objects": 10000
max_size单位是bytes,max_size_kb单位是kb。
操纵radosgw
一般对象存储都由开发在代码层面控制,几乎不需要我们运维人员操作什么,只需要把用户权限和集群维护好就没啥问题了,接下来用一段python代码来演示其bucket以及增删文件的操作。也有一些命令可以实现,例如s3cmd等。
需要先安装好python3环境,以及python的boto模块
# pip3 install boto-2.41.0-py2.py3-none-any.whl
如果没有pip源,离线下载地址:https://pypi.org/simple/boto/
python脚本编写
这里测试使用上边演示新创建的用户vfan
vi ceph-s3.py
import boto.s3.connection
access_key = 'Q6VGP3LYMH99D0A9GUV0' #创建S3用户时返回的AK
secret_key = 'NVDfq7CBJgpUnCXKqbgVuKvI3siWNbx0sRltClA4' #S3用户的SK
host = '10.153.204.13' # RWG节点IP和端口
port = 30080
# 新建一个连接
conn = boto.connect_s3(
aws_access_key_id=access_key,
aws_secret_access_key=secret_key,
host=host, port=port,
is_secure=False, calling_format=boto.s3.connection.OrdinaryCallingFormat(),
)
# 新建一个Bucket
bucket = conn.create_bucket('my-vfan-bucket')
# 列出用户的所有Bucket
for bucket in conn.get_all_buckets():
print("桶名称: %s, 创建时间: %s" %(bucket.name,bucket.creation_date))
# 列出Bucket内容
for key in bucket.list():
print("key名称: %s, 文件大小: %s, 修改时间: %s" %(key.name,key.size,key.last_modified))
# 新建一个对象
key = bucket.new_key('hi.txt')
key.set_contents_from_string('Hello World!')
# 下载一个对象到文件
key = bucket.get_key('hi.txt')
key.get_contents_to_filename('/tmp/hi.txt')
执行py脚本
# python3 ceph-s3.py
桶名称: my-test-bucket, 创建时间: 2022-03-30T10:51:38.321Z
桶名称: my-vfan-bucket, 创建时间: 2022-04-01T07:32:54.671Z
# cat /tmp/hi.txt
Hello World!
已经新创建了一个名为my-vfan-bucket的bucket,并新建了一个对象hi.txt,并下载到了本地的/tmp目录下。
可以优化一下脚本,使其可以单项操作
#!/usr/bin/python
# -*- coding: utf-8 -*-
"""
@Time : 2021-12-22 19:14
@Author : xxxxxx
@Email : xxxxxx
@File : bucket.py
@Software: PyCharm
"""
import boto
import boto.s3.connection
class Bucket():
"""
ceph中bucket相关的类
boto s3 api手册:http://boto.readthedocs.org/en/latest/ref/s3.html
boto s3 api用法:https://docs.ceph.com/en/latest/radosgw/s3/python/#
"""
def __init__(self, ak, sk, host, port):
self.ak = ak
self.sk = sk
self.host = host
self.port = port
self.conn = boto.connect_s3(aws_access_key_id=self.ak, aws_secret_access_key=self.sk, host=self.host,
port=self.port, is_secure=False,
calling_format=boto.s3.connection.OrdinaryCallingFormat())
print self.conn
def bucketList(self):
"""
获取所有的bucketList
:return:
"""
for bucket in self.conn.get_all_buckets():
print("{name}\t{created}".format(name=bucket.name, created=bucket.creation_date))
def bucketCreate(self, bucketName):
"""
创建bucket
:return:
"""
createRes = self.conn.create_bucket(bucketName)
print createRes
def bucketDelete(self):
"""
删除bucket
:return:
"""
pass
if __name__ == "__main__":
"""
主函数
"""
access_key = "FHPC3HED7P7J8ADFQVOD"
secret_key = "Zgf01sjynnAbNS6yCO99VFphDQ6sOlmPBRRd7P2E"
host = "xxxxx"
port = 8000
bucketName = 'share'
bucket = Bucket(access_key, secret_key, host, port)
# 创建bucket
bucket.bucketCreate(bucketName)
# 查看bucket列表
# bucket.bucketList()
可以再基于此脚本优化,增加其他功能。
RadosGW相关操作至此已演示介绍完毕,后续会陆续介绍一些自定义crush规则、pg及一些常用的参数配置。