ceph 004 纠删码池 修改参数 cephx认证
复习ceph003
存储池为逻辑概念，存储池可以占用整个集群的所有空间
[root@ceph01 ~]# ceph osd pool create pool1
pool 'pool1' created
[root@ceph01 ~]# ceph osd pool application enable pool1 rgw
enabled application 'rgw' on pool 'pool1'
[root@ceph01 ~]# ceph osd pool ls detail
pool 1 'device_health_metrics' replicated size 3 min_size 2 crush_rule 0 object_hash rjenkins pg_num 1 pgp_num 1 autoscale_mode on last_change 300 flags hashpspool stripe_width 0 pg_num_min 1 application mgr_devicehealth
pool 4 'pool1' replicated size 3 min_size 2 crush_rule 0 object_hash rjenkins pg_num 32 pgp_num 32 autoscale_mode on last_change 309 flags hashpspool stripe_width 0 application rgw
[root@ceph01 ~]# ceph osd pool ls detail
pool 1 'device_health_metrics' replicated size 3 min_size 2 crush_rule 0 object_hash rjenkins pg_num 1 pgp_num 1 autoscale_mode on last_change 300 flags hashpspool stripe_width 0 pg_num_min 1 application mgr_devicehealth
pool 4 'pool1' replicated size 3 min_size 2 crush_rule 0 object_hash rjenkins pg_num 32 pgp_num 32 autoscale_mode on last_change 309 flags hashpspool stripe_width 0 application rgw
[root@ceph01 ~]# ceph pg dump pgs_brief | grep ^4
dumped pgs_brief
4.1f active+clean [6,5,1] 6 [6,5,1] 6
4.1e active+clean [0,4,7] 0 [0,4,7] 0
4.1d active+clean [7,3,1] 7 [7,3,1] 7
4.1c active+clean [6,1,4] 6 [6,1,4] 6
4.1b active+clean [4,1,8] 4 [4,1,8] 4
4.1a active+clean [4,0,7] 4 [4,0,7] 4
4.19 active+clean [6,0,5] 6 [6,0,5] 6
4.18 active+clean [3,1,6] 3 [3,1,6] 3
4.17 active+clean [0,7,3] 0 [0,7,3] 0
4.16 active+clean [0,4,8] 0 [0,4,8] 0
4.15 active+clean [8,0,5] 8 [8,0,5] 8
4.14 active+clean [1,5,7] 1 [1,5,7] 1
4.13 active+clean [4,6,1] 4 [4,6,1] 4
4.12 active+clean [1,8,4] 1 [1,8,4] 1
4.11 active+clean [4,0,7] 4 [4,0,7] 4
4.10 active+clean [1,6,4] 1 [1,6,4] 1
4.f active+clean [7,1,5] 7 [7,1,5] 7
4.4 active+clean [1,7,3] 1 [1,7,3] 1
4.3 active+clean [6,5,1] 6 [6,5,1] 6
4.2 active+clean [5,6,2] 5 [5,6,2] 5
4.1 active+clean [6,5,0] 6 [6,5,0] 6
4.5 active+clean [3,6,0] 3 [3,6,0] 3
4.0 active+clean [0,4,6] 0 [0,4,6] 0
4.6 active+clean [1,4,6] 1 [1,4,6] 1
4.7 active+clean [1,6,3] 1 [1,6,3] 1
4.8 active+clean [8,3,0] 8 [8,3,0] 8
4.9 active+clean [6,1,5] 6 [6,1,5] 6
4.a active+clean [5,1,7] 5 [5,1,7] 5
4.b active+clean [8,1,4] 8 [8,1,4] 8
4.c active+clean [4,1,8] 4 [4,1,8] 4
4.d active+clean [7,1,4] 7 [7,1,4] 7
4.e active+clean [4,6,1] 4 [4,6,1] 4
先写主后写备。三副本有自我恢复机制，在主osd坏掉后，备暂时成为主。然后再找一个osd成为第三个副本
老版不会自动扩
pg自动扩大 从 32开始
pg计算器
100osd 每个osd承载pg数量有限（100-200 pg/osd）
100osd * 200 = 20000pg
规划创建2个存储池
pool1 10000pg
pool2 10000pg
这个池osd承载数量可以改，但最好不要
在这个20000pg平均下来，性能会好一些
rados为集群内部的操作命令，排错调试
mom_allow_pool_delete = true全局图像化修改此参数允许删除存储池
ceph osd pool set pool1 nodelete false 单独设置存储池可不可删除
存储池命名空间
分隔存储池中数据，允许访问哪一个存储池里面的哪一个命名空间
[root@ceph01 ~]# cp /etc/passwd .
[root@ceph01 ~]#
[root@ceph01 ~]# rados -p pool1 put password passwd
[root@ceph01 ~]# rados -p pool1 ls
password
[root@ceph01 ~]# rados -p pool1 -N sys put password01 passwd
[root@ceph01 ~]# rados -p pool1 ls
password
[root@ceph01 ~]# rados -p pool1 -N sys ls
password01
[root@ceph01 ~]# rados -p pool1 --all ls
password
sys password01
允许用户访问存储池，就可以访问所有的命名空间
只限制用户只能访问 sys这个命名空间，其他命名空间就无法访问
纠删码池
10M 复制池 30M
10M 纠删码 < 30M 纠删码池可以节省空间
n = k + m
4M
k为数据块 2 每个数据块大小2M 如果为3数据块则为1.3 1.3*3
m为编码块 2 编码块2M 1.3*2 1.3*5=6.5M
两个数据块基于算法，会得到两个校验块 （raid6）
10M占了20M
pool 5 'pool2' erasure profile default size 4 min_size 3 crush_rule 1 object_hash rjenkins pg_num 32 pgp_num 32 autoscale_mode on last_change 313 flags hashpspool stripe_width 8192
[root@ceph01 ~]# ceph osd erasure-code-profile get default
k=2
m=2
plugin=jerasure
technique=reed_sol_van
[root@ceph01 ~]#
四个副本最多只能坏一个副本
坏2校验块数据不丢，但谁能保证坏的两个都是校验快呢
[root@ceph01 ~]# ceph osd pool application enable pool2 rgw
enabled application 'rgw' on pool 'pool2'
[root@ceph01 ~]# rados -p pool2 put passwd02 passwd
[root@ceph01 ~]# ceph pg dump pgs_brief | grep ^5
dumped pgs_brief
5.f active+undersized [5,2147483647,7,0] 5 [5,2147483647,7,0] 5
5.c active+undersized [1,3,6,2147483647] 1 [1,3,6,2147483647] 1
5.d active+undersized [7,1,3,2147483647] 7 [7,1,3,2147483647] 7
5.a active+undersized [2,6,2147483647,5] 2 [2,6,2147483647,5] 2
5.b active+undersized [7,2,4,2147483647] 7 [7,2,4,2147483647] 7
5.8 active+undersized [6,1,2147483647,4] 6 [6,1,2147483647,4] 6
5.9 active+undersized [7,5,2147483647,2] 7 [7,5,2147483647,2] 7
5.6 active+undersized [2,4,8,2147483647] 2 [2,4,8,2147483647] 2
5.7 active+undersized [5,0,7,2147483647] 5 [5,0,7,2147483647] 5
5.1 active+undersized [3,0,2147483647,7] 3 [3,0,2147483647,7] 3
5.4 active+undersized [7,2,4,2147483647] 7 [7,2,4,2147483647] 7
5.0 active+undersized [6,2,3,2147483647] 6 [6,2,3,2147483647] 6
5.3 active+undersized [0,2147483647,4,6] 0 [0,2147483647,4,6] 0
5.2 active+undersized [5,2,7,2147483647] 5 [5,2,7,2147483647] 5
5.5 active+undersized [0,7,4,2147483647] 0 [0,7,4,2147483647] 0
5.e active+undersized [6,4,1,2147483647] 6 [6,4,1,2147483647] 6
5.11 active+undersized [6,3,2147483647,1] 6 [6,3,2147483647,1] 6
5.10 active+undersized [4,6,2,2147483647] 4 [4,6,2,2147483647] 4
5.13 active+undersized [6,1,2147483647,4] 6 [6,1,2147483647,4] 6
5.12 active+undersized [6,3,2,2147483647] 6 [6,3,2,2147483647] 6
5.15 active+undersized [5,0,8,2147483647] 5 [5,0,8,2147483647] 5
5.14 active+undersized [0,7,4,2147483647] 0 [0,7,4,2147483647] 0
5.17 active+undersized [2,8,2147483647,4] 2 [2,8,2147483647,4] 2
5.16 active+undersized [3,1,2147483647,8] 3 [3,1,2147483647,8] 3
5.19 active+undersized [1,5,2147483647,7] 1 [1,5,2147483647,7] 1
5.18 active+undersized [4,7,1,2147483647] 4 [4,7,1,2147483647] 4
5.1b active+undersized [5,1,2147483647,7] 5 [5,1,2147483647,7] 5
5.1a active+undersized [8,4,0,2147483647] 8 [8,4,0,2147483647] 8
5.1d active+undersized [1,6,5,2147483647] 1 [1,6,5,2147483647] 1
5.1c active+undersized [4,6,0,2147483647] 4 [4,6,0,2147483647] 4
5.1f active+undersized [3,2147483647,0,8] 3 [3,2147483647,0,8] 3
5.1e active+undersized [0,7,3,2147483647] 0 [0,7,3,2147483647] 0
[root@ceph01 ~]# ceph osd map pool2 passwd02
osdmap e314 pool 'pool2' (5) object 'passwd02' -> pg 5.e793df76 (5.16) -> up ([3,1,NONE,8], p3) acting ([3,1,NONE,8], p3)
[root@ceph01 ~]#
更改故障域host 为 osd
[root@ceph01 ~]# ceph osd erasure-code-profile set ec01 crush-failure-domain=osd k=3 m=2
[root@ceph01 ~]# ceph osd erasure-code-profile get ec01
crush-device-class=
crush-failure-domain=osd
crush-root=default
jerasure-per-chunk-alignment=false
k=3
m=2
plugin=jerasure
technique=reed_sol_van
w=8
[root@ceph01 ~]# ceph osd pool create pool3 erasure ec01
pool 'pool3' created
[root@ceph01 ~]# ceph pg dump pgs_brief | grep ^6
dumped pgs_brief
6.1d active+clean [1,8,6,2,5] 1 [1,8,6,2,5] 1
6.1c active+clean [7,1,3,4,2] 7 [7,1,3,4,2] 7
6.1f active+clean [2,6,8,7,3] 2 [2,6,8,7,3] 2
6.1e active+clean [8,6,5,2,4] 8 [8,6,5,2,4] 8
6.19 active+clean [5,1,8,6,0] 5 [5,1,8,6,0] 5
6.18 active+clean [2,1,0,6,3] 2 [2,1,0,6,3] 2
6.1b active+clean [6,8,7,5,3] 6 [6,8,7,5,3] 6
6.1a active+clean [4,3,1,8,6] 4 [4,3,1,8,6] 4
6.15 active+clean [3,0,8,2,7] 3 [3,0,8,2,7] 3
6.14 active+clean [2,4,3,5,8] 2 [2,4,3,5,8] 2
6.17 active+clean [6,2,8,7,0] 6 [6,2,8,7,0] 6
6.16 active+clean [0,4,2,3,5] 0 [0,4,2,3,5] 0
6.11 active+clean [2,0,8,6,7] 2 [2,0,8,6,7] 2
6.10 active+clean [0,8,1,5,7] 0 [0,8,1,5,7] 0
6.13 active+clean [2,1,8,3,4] 2 [2,1,8,3,4] 2
6.12 active+clean [8,7,6,2,3] 8 [8,7,6,2,3] 8
6.d active+clean [5,1,8,0,7] 5 [5,1,8,0,7] 5
6.6 active+clean [1,4,3,8,0] 1 [1,4,3,8,0] 1
6.1 active+clean [6,4,1,3,5] 6 [6,4,1,3,5] 6
6.0 active+clean [0,7,6,5,3] 0 [0,7,6,5,3] 0
6.3 active+clean [4,5,7,3,0] 4 [4,5,7,3,0] 4
6.7 active+clean [5,2,0,4,1] 5 [5,2,0,4,1] 5
6.2 active+clean [3,0,2,8,5] 3 [3,0,2,8,5] 3
6.4 active+clean [1,5,6,2,3] 1 [1,5,6,2,3] 1
6.5 active+clean [5,4,1,2,3] 5 [5,4,1,2,3] 5
6.a active+clean [5,4,0,7,2] 5 [5,4,0,7,2] 5
6.b active+clean [1,3,4,6,7] 1 [1,3,4,6,7] 1
6.8 active+clean [3,6,2,8,4] 3 [3,6,2,8,4] 3
6.9 active+clean [0,5,7,2,6] 0 [0,5,7,2,6] 0
6.e active+clean [4,8,1,2,5] 4 [4,8,1,2,5] 4
6.f active+clean [2,6,7,3,4] 2 [2,6,7,3,4] 2
6.c active+clean [1,4,5,8,3] 1 [1,4,5,8,3] 1
4M的对象你切分成3个数据块，
数据发生变化也要重新计算
k=3 m=2 就最多坏2个 3+2好一些 但是最多让你坏1个保险
创建完ec01规则不能修改 如果有存储池在应用这个规则
性能比复制池差，因为时通过校验块，来保证可靠性，所以会消耗cpu计算
一个文件被切成对象，如果用的是纠删码池，对象还得被切块，就是/k的数量，然后计算出编码块来保证可靠性。
一些相关的参数
k m rule(crush-failure-domain)
crush-device-class 仅将某一类设备支持的osd用于池 ssd hdd nvme
crush-root 设置crush规则集的根节点
ceph的配置
/etc/ceph
ceph.conf 集群配置文件的入口
[root@ceph01 ceph]# cat ceph.conf
# minimal ceph.conf for cb8f4abe-14a7-11ed-a76d-000c2939fb75
[global]
fsid = cb8f4abe-14a7-11ed-a76d-000c2939fb75
mon_host = [v2:192.168.92.11:3300/0,v1:192.168.92.11:6789/0]
mon其实有三个，可以手动更新一下
更新ceph配置
集群网络（心跳，重平衡之类），和客户端访问的网络得分开好一些
[root@ceph01 ceph]# ceph config get mon public_network
192.168.92.0/24
[root@ceph01 ceph]# ceph config get mon cluster_network
xxxxxx
我没设置
可以通过set改
[root@ceph01 ceph]# ceph -s --cluster ceph
cluster:
id: cb8f4abe-14a7-11ed-a76d-000c2939fb75
health: HEALTH_WARN
1 osds exist in the crush map but not in the osdmap
Degraded data redundancy: 1/10 objects degraded (10.000%), 1 pg degraded, 32 pgs undersized
services:
mon: 3 daemons, quorum ceph01.example.com,ceph02,ceph03 (age 22h)
mgr: ceph01.example.com.wvuoii(active, since 22h), standbys: ceph02.alqzfq
osd: 9 osds: 9 up (since 22h), 9 in (since 2d)
data:
pools: 4 pools, 97 pgs
objects: 3 objects, 2.9 KiB
usage: 9.4 GiB used, 171 GiB / 180 GiB avail
pgs: 1/10 objects degraded (10.000%)
65 active+clean
31 active+undersized
1 active+undersized+degraded
[root@ceph01 ceph]# ls
COPYING sample.ceph.conf
[root@ceph01 ceph]# pwd
/usr/share/doc/ceph
[root@ceph01 ceph]# cd /etc/ceph/
[root@ceph01 ceph]# ls
ceph.client.admin.keyring ceph.conf ceph.pub rbdmap
[root@ceph01 ceph]#
ceph.conf .conf前面的就是集群名
区分集群名
ceph -s
health: HEALTH_WARN
1 osds exist in the crush map but not in the osdmap
Degraded data redundancy: 1/10 objects degraded (10.000%), 1 pg degraded, 32 pgs undersized
纠删码那个池引发的报错
[root@ceph01 ceph]# ceph osd pool delete pool2 pool2 --yes-i-really-really-mean-it
Error EPERM: pool deletion is disabled; you must first set the mon_allow_pool_delete config option to true before you can destroy a pool
[root@ceph01 ceph]# ceph config set mon mon_allow_pool_delete true
[root@ceph01 ceph]# ceph osd pool delete pool2 pool2 --yes-i-really-really-mean-it
pool 'pool2' removed
[root@ceph01 ceph]#
[root@ceph01 ceph]# ceph -s
cluster:
id: cb8f4abe-14a7-11ed-a76d-000c2939fb75
health: HEALTH_WARN
1 osds exist in the crush map but not in the osdmap
services:
mon: 3 daemons, quorum ceph01.example.com,ceph02,ceph03 (age 23h)
mgr: ceph01.example.com.wvuoii(active, since 23h), standbys: ceph02.alqzfq
osd: 9 osds: 9 up (since 23h), 9 in (since 2d)
data:
health_warn一般问题不大
报错信息就会少一条
ceph -s --name client.admin
ceph -s --id admin
元数据用来区分
在客户端区分做区分集群名之类
配置参数
以前改配置文件，现在通过命令行改
图形界面也可以看
[root@ceph01 ceph]# ceph config ls | grep mon | grep delete
mon_allow_pool_delete
mon_fake_pool_delete
[root@ceph01 ceph]# ceph config get mon mon_allow_pool_delete
true
对存储池操作由mon
对osd读写由osd决定
写在数据库，立即生效
ceph orch restart mon 重启集群所有mon
ceph orch daemon restart ods.1 针对性重启
ceph config dump
查看数据库的当前值
[root@ceph01 ~]# ceph config ls | grep mon | grep delete
mon_allow_pool_delete
mon_fake_pool_delete
过滤有用的设置值
[root@ceph01 ~]# ceph config set mon mon_allow_pool_delete false
[root@ceph01 ~]# ceph config get mon mon_allow_pool_delete
false
更改值并查看
或者
写一个配置文件
[root@ceph01 ~]# cat ceph.conf
[global]
mon_allow_pool_delete = false
使用此配置文件修改值
使用这个文件生效
ceph config assimilate-conf -i ceph.conf
可以知道自己写了哪些内容
ceph 认证授权
ceph auth ls 查看所有用户
[root@serverc ~]# ceph config ls | grep auth | grep required
auth_cluster_required
auth_service_required
auth_client_required
[root@serverc ~]# ceph config get mon auth_cluster_required
cephx
创建密钥
key来认证，然后访问集群
访问
创建用户
[root@ceph01 ~]# ceph auth get-or-create client.user1 --id admin
[client.user1]
key = AQB+PPdiAgZZBxAAWKZ3s0Cu3y7o7yEVZqgKKQ==
使用id admin创建的
[root@ceph01 ~]# ceph auth get-or-create client.admin > /etc/ceph/ceph.client.user1.keyring
[root@ceph01 ~]# ceph -s --name client.user1
2022-08-13T13:56:00.302+0800 7f6eaaffd700 -1 monclient(hunting): handle_auth_bad_method server allowed_methods [2] but i only support [2,1]
[errno 13] RADOS permission denied (error connecting to the cluster)
[root@ceph01 ~]#
无权限
权限：
r：mon 获取集群信息
osd 获取对象信息
w：mon 修改集群信息，mon参数，创建和修改存储池
osd 上传对象 删除对象
授权
[root@serverc ~]# ceph auth caps client.user1 mon 'allow r'
updated caps for client.user1
[root@serverc ~]# ceph osd pool create pool1 --id user1
Error EACCES: access denied
[root@serverc ~]# ceph auth caps client.user1 mon 'allow rw'
updated caps for client.user1
[root@serverc ~]# ceph osd pool create pool1 --id user1
pool 'pool1' created
[root@serverc ~]#
[root@serverc ceph]# ceph auth caps client.user1 mon 'allow rw' osd 'allow r'
updated caps for client.user1
[root@serverc ceph]# rados -p pool1 ls --id user1
[root@serverc ceph]#
[root@serverc ceph]# ceph auth caps client.user1 mon 'allow rw' osd 'allow rw'
updated caps for client.user1
[root@serverc ceph]# rados -p pool1 ls --id user1
[root@serverc ceph]# cp /etc/passwd .
[root@serverc ceph]# rados -p pool1 put file2 passwd --id user1
[root@serverc ceph]# rados -p pool1 ls --id user1
file2
查看权限
[root@ceph01 ceph]# ceph auth get client.user1
exported keyring for client.user1
[client.user1]
key = AQB+PPdiAgZZBxAAWKZ3s0Cu3y7o7yEVZqgKKQ==
caps mon = "allow r"
[root@ceph01 ceph]#
