使用 Ceph RBD 做为 kubernetes 后端存储
Ceph 安装部署
由于这里我们使用 RBD 所以我们使用到的组件为 Ceph.mon, Ceph.osd, 这两个组件就可以了。 Ceph.mds 为 cephfs 所需组件
部署环境
1
2
3
4
5
6
7
8
| # Ceph.Mon = 2n+1 个 ，3个的情况下只能掉线一个，如果同时2个掉线
# 集群会出现无法仲裁，集群会一直等待 Ceph.Mon 恢复超过半数。
172.16.1.37 Ceph.admin + Ceph.Mon
172.16.1.38 Ceph.Mon + Ceph.osd
172.16.1.39 Ceph.Mon + Ceph.osd
172.16.1.40 Ceph.osd
|
初始化环境
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
| # 1. ceph 对时间要求很严格， 一定要同步所有的服务器时间
# 2. 配置 hosts
cat << "EOF" >> /etc/hosts
# --- Ceph Hosts -----
172.16.1.37 ceph-node-1
172.16.1.38 ceph-node-2
172.16.1.39 ceph-node-3
172.16.1.40 ceph-node-4
# --- Ceph Hosts -----
EOF
# 3. 配置 hostname
hostnamectl --static set-hostname ceph-node-1
hostnamectl --static set-hostname ceph-node-2
hostnamectl --static set-hostname ceph-node-3
hostnamectl --static set-hostname ceph-node-4
# 4. Ceph.admin 节点 配置无密码ssh
ssh-keygen
ssh-copy-id ceph-node-1 -p33
ssh-copy-id ceph-node-2 -p33
ssh-copy-id ceph-node-3 -p33
ssh-copy-id ceph-node-4 -p33
Now try logging into the machine, with: "ssh -p '33' 'ceph-node-2'"
# 5. 增加 ~/.ssh/config 文件, 否则 ceph-deploy 创建集群 报端口错误
[root@ceph-node-1 ~]# vi ~/.ssh/config
Host ceph-node-1
Hostname ceph-node-1
Port 33
Host ceph-node-2
Hostname ceph-node-2
Port 33
Host ceph-node-3
Hostname ceph-node-3
Port 33
Host ceph-node-4
Hostname ceph-node-4
Port 33
|
安装 Ceph-deploy
1
2
3
4
5
6
7
8
9
10
11
12
13
14
| # 管理节点 安装 ceph-deploy 管理工具
# 配置 官方 的 Ceph 源
rpm --import https://download.ceph.com/keys/release.asc
rpm -Uvh --replacepkgs https://download.ceph.com/rpm-jewel/el7/noarch/ceph-release-1-0.el7.noarch.rpm
# 安装 epel 源
rpm -Uvh http://mirrors.ustc.edu.cn/centos/7/extras/x86_64/Packages/epel-release-7-9.noarch.rpm
[root@ceph-node-1 ~]# yum makecache
[root@ceph-node-1 ~]# yum -y install ceph-deploy ntpdate
|
创建 Ceph-Mon
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
| # 创建集群目录，用于存放配置文件，证书等信息
mkdir -p /opt/ceph-cluster
cd /opt/ceph-cluster/
# 创建ceph-mon 节点
[root@ceph-node-1 ceph-cluster]# ceph-deploy new ceph-node-1 ceph-node-2 ceph-node-3
Monitor initial members are ['ceph-node-1', 'ceph-node-2', 'ceph-node-3']
# 查看配置文件
[root@ceph-node-1 ceph-cluster]# cat ceph.conf
[global]
fsid = 211a5d26-0377-4d1c-8555-c4793cef83c1
mon_initial_members = ceph-node-1, ceph-node-2, ceph-node-3
mon_host = 172.16.1.37,172.16.1.38,172.16.1.39
auth_cluster_required = cephx
auth_service_required = cephx
auth_client_required = cephx
# 修改 osd 的副本数，既数据保存N份。
echo 'osd_pool_default_size = 2' >> ./ceph.conf
# 注: 如果文件系统为 ext4 请添加
echo 'osd max object name len = 256' >> ./ceph.conf
echo 'osd max object namespace len = 64' >> ./ceph.conf
|
安装 Ceph
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
| # 可通过 ceph-deploy 安装，也可以登陆node 本地安装
# 命令: ceph-deploy install {ceph-node}[{ceph-node} ...]
[root@ceph-node-1 ceph-cluster]# ceph-deploy install ceph-node-1 ceph-node-2 ceph-node-3 ceph-node-4
# 使用 cpeh-deploy 安装，会一直使用官方源下载，超级慢
# 如果出现超时，请自行在每个服务器中使用 yum 安装
# 替换 ceph 源 为 163 源
sed -i 's/download\.ceph\.com/mirrors\.163\.com\/ceph/g' /etc/yum.repos.d/ceph.repo
yum -y install ceph ceph-radosgw
# 检测 安装
[root@ceph-node-1 ceph-cluster]# ceph --version
ceph version 10.2.7 (50e863e0f4bc8f4b9e31156de690d765af245185)
|
初始化 ceph-mon 节点
1
2
3
4
5
6
7
8
9
10
11
12
13
| # 请务必在 ceph-cluster 目录下
[root@ceph-node-1 ceph-cluster]# ceph-deploy mon create-initial
[ceph_deploy.gatherkeys][INFO ] Storing ceph.client.admin.keyring
[ceph_deploy.gatherkeys][INFO ] Storing ceph.bootstrap-mds.keyring
[ceph_deploy.gatherkeys][INFO ] keyring 'ceph.mon.keyring' already exists
[ceph_deploy.gatherkeys][INFO ] Storing ceph.bootstrap-osd.keyring
[ceph_deploy.gatherkeys][INFO ] Storing ceph.bootstrap-rgw.keyring
[ceph_deploy.gatherkeys][INFO ] Destroy temp directory /tmp/tmpNfXVDA
|
初始化 ceph.osd 节点
1
2
3
4
5
6
7
8
9
10
11
12
13
| # 首先创建 存储空间, 如果使用分区，可略过
[root@ceph-node-2 ~]# mkdir -p /opt/ceph-osd-1 && chown ceph:ceph /opt/ceph-osd-1
[root@ceph-node-3 ~]# mkdir -p /opt/ceph-osd-2 && chown ceph:ceph /opt/ceph-osd-2
[root@ceph-node-4 ~]# mkdir -p /opt/ceph-osd-3 && chown ceph:ceph /opt/ceph-osd-3
# 启动 osd
[root@ceph-node-1 ceph-cluster]# ceph-deploy osd prepare ceph-node-2:/opt/ceph-osd-1 ceph-node-3:/opt/ceph-osd-2 ceph-node-4:/opt/ceph-osd-3
|
1
2
3
| # 激活 所有 osd 节点
[root@ceph-node-1 ceph-cluster]# ceph-deploy osd activate ceph-node-2:/opt/ceph-osd-1 ceph-node-3:/opt/ceph-osd-2 ceph-node-4:/opt/ceph-osd-3
|
1
2
| # 把管理节点的配置文件与keyring同步至其它节点
[root@ceph-node-1 ceph-cluster]# ceph-deploy admin ceph-node-1 ceph-node-2 ceph-node-3 ceph-node-4
|
1
2
3
4
5
6
7
8
9
10
11
12
13
| # 没有出现 红色的 ERROR
# 查看状态
[root@ceph-node-1 ceph-cluster]# ceph -s
cluster 7ab7e076-24e8-4236-bd4e-88675aab7365
health HEALTH_OK
monmap e1: 3 mons at {ceph-node-1=172.16.1.37:6789/0,ceph-node-2=172.16.1.38:6789/0,ceph-node-3=172.16.1.39:6789/0}
election epoch 6, quorum 0,1,2 ceph-node-1,ceph-node-2,ceph-node-3
osdmap e15: 3 osds: 3 up, 3 in
flags sortbitwise,require_jewel_osds
pgmap v24: 64 pgs, 1 pools, 0 bytes data, 0 objects
15628 MB used, 8957 GB / 9452 GB avail
64 active+clean
|
1
2
3
4
5
6
7
8
9
10
11
12
| # 查看 osd tree
[root@ceph-node-1 ceph-cluster]# ceph osd tree
ID WEIGHT TYPE NAME UP/DOWN REWEIGHT PRIMARY-AFFINITY
-1 9.23126 root default
-2 3.07709 host ceph-node-2
0 3.07709 osd.0 up 1.00000 1.00000
-3 3.07709 host ceph-node-3
1 3.07709 osd.1 up 1.00000 1.00000
-4 3.07709 host ceph-node-4
2 3.07709 osd.2 up 1.00000 1.00000
|
1
2
3
4
5
6
| # 设置所有开机启动
systemctl enable ceph-osd.target
systemctl enable ceph-mon.target
|
1
2
3
4
5
6
7
8
| # 重启系统以后重启 osd
#查看处于down 的 osd
ceph osd tree
# 登陆所在node 启动ceph-osd进程 [id 为 tree 查看的 id]
systemctl start ceph-osd@id
|
kubernetes Volume Ceph RBD
官方 RDB 的文件 https://github.com/kubernetes/kubernetes/tree/master/examples/volumes/rbd
k8s 安装 ceph-common
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
| # 1. 在所有的 k8s node 里面安装 ceph-common
yum -y install ceph-common
# 2. 拷贝 ceph.conf 与 ceph.client.admin.keyring
拷贝到 /etc/ceph/ 目录下
# 3. 配置 kubelet 增加ceph参数
# 这里最好是在创建 k8s 集群的时候就添加好
vim /usr/local/bin/kubelet
# 增加 如下
-v /etc/ceph:/etc/ceph:ro \
-v /sbin/modprobe:/sbin/modprobe:ro \
-v /usr/sbin/modprobe:/usr/sbin/modprobe:ro \
-v /lib/modules:/lib/modules:ro \
# 重启 kubelet
systemctl restart kubelet.service
# 否则创建 pod 的时候 报错
with: rbd: failed to modprobe rbd error:exit status 1
|
修改 ceph 配置
1
2
3
4
5
6
7
8
9
10
11
12
| # 修改配置
# rbd image有4个 features，layering, exclusive-lock, object-map, fast-diff, deep-flatten
因为目前内核仅支持layering，修改默认配置。
[root@ceph-node-1 ceph-cluster]# echo 'rbd_default_features = 1' >> ./ceph.conf
# 验证一下
[root@ceph-node-1 ceph-cluster]# ceph --show-config|grep rbd|grep features
rbd_default_features = 1
# 把管理节点的配置文件与keyring同步至其它节点
[root@ceph-node-1 ceph-cluster]# ceph-deploy --overwrite-conf admin ceph-node-1 ceph-node-2 ceph-node-3 ceph-node-4
|
创建 ceph-secret
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
| # 获取 client.admin 的值
[root@ceph-node-1 ceph-cluster]# ceph auth get-key client.admin
AQBpUAxZGmDBBxAAVbgeRss9jv39dE0biTE7qQ==
# 转换成 base64 编码
[root@ceph-node-1 ceph-cluster]# echo "AQBpUAxZGmDBBxAAVbgeRss9jv39dE0biTE7qQ=="|base64
QVFCcFVBeFpHbURCQnhBQVZiZ2VSc3M5anYzOWRFMGJpVEU3cVE9PQo=
# 创建ceph-secret.yaml文件
apiVersion: v1
kind: Secret
metadata:
name: ceph-secret
data:
key: QVFCcFVBeFpHbURCQnhBQVZiZ2VSc3M5anYzOWRFMGJpVEU3cVE9PQo=
# 导入 yaml 文件
[root@k8s-node-28 cephrbd]# kubectl apply -f ceph-secret.yaml
secret "ceph-secret" created
# 查看 状态
[root@k8s-node-28 cephrbd]# kubectl get secret |grep ceph
ceph-secret Opaque 1 53s
|
创建 ceph image
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
| # 创建一个 1G 的 image
[root@ceph-node-1 ceph-cluster]# rbd create test-image -s 1G
# 查看 image
[root@ceph-node-1 ceph-cluster]# rbd list
test-image
[root@ceph-node-1 ceph-cluster]# rbd info test-image
rbd image 'test-image':
size 1024 MB in 256 objects
order 22 (4096 kB objects)
block_name_prefix: rbd_data.373b2ae8944a
format: 2
features: layering
flags:
# 这里 format 为 2 , 如果旧系统 不支持 format 2 ，可将 format 设置为 1 。
|
创建一个 pv
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
| # 创建 test-pv.yaml
apiVersion: v1
kind: PersistentVolume
metadata:
name: test-pv
spec:
capacity:
storage: 1Gi
accessModes:
- ReadWriteOnce
rbd:
monitors:
- 172.16.1.37:6789
- 172.16.1.38:6789
- 172.16.1.38:6789
pool: rbd
image: test-image
user: admin
secretRef:
name: ceph-secret
fsType: ext4
readOnly: false
persistentVolumeReclaimPolicy: Recycle
# 导入 yaml 文件
[root@k8s-node-28 cephrbd]# kubectl apply -f test-pv.yaml
persistentvolume "test-pv" created
# 查看 pv
[root@k8s-node-28 cephrbd]# kubectl get pv |grep test-pv
test-pv 1Gi RWO Recycle Available 33s
|
创建一个 pvc
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
| # vi test-pvc.yaml
kind: PersistentVolumeClaim
apiVersion: v1
metadata:
name: test-claim
spec:
accessModes:
- ReadWriteOnce
resources:
requests:
storage: 1Gi
# 导入 yaml 文件
[root@k8s-node-28 cephrbd]# kubectl apply -f test-pvc.yaml
persistentvolumeclaim "test-claim" created
# 查看 pvc
[root@k8s-node-28 cephrbd]# kubectl get pvc |grep test
test-claim Bound test-pv 1Gi RWO 31s
|
创建一个 deployment
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
| vi nginx-deplyment.yaml
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
name: nginx-dm
spec:
replicas: 1
template:
metadata:
labels:
name: nginx
spec:
containers:
- name: nginx
image: nginx:alpine
imagePullPolicy: IfNotPresent
ports:
- containerPort: 80
volumeMounts:
- name: ceph-rbd-volume
mountPath: "/usr/share/nginx/html"
volumes:
- name: ceph-rbd-volume
persistentVolumeClaim:
claimName: test-claim
# 导入 yaml 文件
[root@k8s-node-28 cephrbd]# kubectl apply -f nginx-deplyment.yaml
deployment "nginx-dm" configured
# 查看 pods
[root@k8s-node-28 cephrbd]# kubectl get pods
NAME READY STATUS RESTARTS AGE
nginx-dm-1697116629-j4m1g 1/1 Running 0 8m
|
测试
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
| # 查看 分区文件
[root@k8s-node-28 cephrbd]# kubectl exec -it nginx-dm-1697116629-j4m1g -- df -h
Filesystem Size Used Available Use% Mounted on
overlay 115.2G 4.9G 104.4G 4% /
tmpfs 62.8G 0 62.8G 0% /dev
tmpfs 62.8G 0 62.8G 0% /sys/fs/cgroup
/dev/mapper/centos-root
115.2G 4.9G 104.4G 4% /dev/termination-log
/dev/mapper/centos-root
115.2G 4.9G 104.4G 4% /etc/resolv.conf
/dev/mapper/centos-root
115.2G 4.9G 104.4G 4% /etc/hostname
/dev/mapper/centos-root
115.2G 4.9G 104.4G 4% /etc/hosts
shm 64.0M 0 64.0M 0% /dev/shm
/dev/rbd0 975.9M 1.3M 907.4M 0% /usr/share/nginx/html
tmpfs 62.8G 12.0K 62.8G 0% /var/run/secrets/kubernetes.io/serviceaccount
tmpfs 62.8G 0 62.8G 0% /proc/kcore
tmpfs 62.8G 0 62.8G 0% /proc/timer_list
tmpfs 62.8G 0 62.8G 0% /proc/timer_stats
tmpfs 62.8G 0 62.8G 0% /proc/sched_debug
tmpfs 62.8G 0 62.8G 0% /sys/firmware
## 写入测试
[root@k8s-node-28 cephrbd]# kubectl exec -it nginx-dm-1697116629-j4m1g -- touch /usr/share/nginx/html/jicki.html
# 删除 pod
[root@k8s-node-28 cephrbd]# kubectl delete pod/nginx-dm-1697116629-j4m1g
pod "nginx-dm-1697116629-j4m1g" deleted
# 新的 pod
[root@k8s-node-28 cephrbd]# kubectl get pods
NAME READY STATUS RESTARTS AGE
nginx-dm-1697116629-v2n52 1/1 Running 0 11s
# 查看文件
[root@k8s-node-28 cephrbd]# kubectl exec -it nginx-dm-1697116629-v2n52 -- ls -lt /usr/share/nginx/html
total 16
-rw-r--r-- 1 root root 0 May 9 07:53 jicki.html
|
FAQ and Error
FAQ 1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
| # ceph 挂载关于 namespace != default 的情况下
1. 需要配置 secret = namespace
kubectl get secret 可以看到只存在 default 的 认证
2. 需要配置 pv = namespace ，虽然pv 是全局的，但是最好还是配置一下
kubectl get pv --namespace=
3. 需要配置 pvc = namespace 。
kubectl get pvc --namespace=
# 确认以上3点，否则在挂载namespace 的时候会报错
ceph timeout expired waiting for volumes to attach/mount for pod
|
Error 1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
| # 删除 osd 如果创建出错，请删除 osd
# 使用 ceph osd tree 查看出错的 osd.id
ceph osd out id 提出集群
# 登陆所在 服务器 停止服务
systemctl stop ceph-osd@id
# 然后再执行 crush remove 删除 tree
ceph osd crush remove osd.id
# 最后 执行 auth del 和 rm 删除
ceph auth del osd.id
ceph osd rm id
# 在运行 tree 已经不存在了
# 如果使用分区挂载 还需要 登陆 osd 节点中 umount 分区
# 使用 lsblk 查看挂载
# 如: /dev/sdb 被挂载
umount /dev/sdb
ceph-disk zap /dev/sdb
|
Error 2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
| # 如果暴力删除了 osd 然后重建了 osd
ceph -s
报错 stale+active+undersized+degraded+remapped
# 执行 ceph health detail 查看所有报错的信息
# 执行 ceph pg <pgid> query 查看具体信息
# 如果执行 query 报错
# 使用 ceph pg force_create_pg <pgid> 覆盖错误
# 如果 datail 有很多 pgid 出错 使用 for 循环去跑
for pg in `ceph health detail | grep "stale+active+undersized+degraded" | awk '{print $2}' | sort | uniq`;
do
ceph pg force_create_pg $pg
done
# 如果仍然不能解决问题，那么请使用暴力方法
# 重建ceph 集群 删除集群所有文件,重新部署
systemctl stop ceph-osd.target
systemctl stop ceph-mon.target
rm -rf /etc/ceph/*
rm -rf /var/log/ceph/*
rm -rf /var/lib/ceph/*
# osd 清除挂载 目录的里的所有内容
# 卸载 ceph
yum -y remove ceph ceph-radosgw
# 重新安装
yum -y install ceph ceph-radosgw
# 在 osd 节点 创建目录
mkdir -p /var/lib/ceph/osd/
|
无法删除 rbd image
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
| # 删除报错
[root@ceph-node-1 ~]# rbd rm test-image
2017-07-07 11:41:39.733710 7f7a55898d80 -1 librbd: image has watchers - not removing
Removing image: 0% complete...failed.
rbd: error: image still has watchers
This means the image is still open or the client using it crashed. Try again after closing/unmapping it or waiting 30s for the crashed client to timeout.
# 查看 状态 与 客户端连接
[root@ceph-node-1 ~]# rbd status test-image
Watchers:
watcher=172.16.1.29:0/1324761697 client.14212 cookie=1
watcher=172.16.1.42:0/3859462470 client.14226 cookie=1
# 登陆所在 服务器 29
[root@k8s-node-29 ~]# rbd showmapped
id pool image snap device
0 rbd test-image - /dev/rbd0
# 执行 rbd unmap
[root@k8s-node-29 ~]# rbd unmap /dev/rbd0
# 登陆所在 服务器 42
[root@k8s-node-42 ~]# rbd showmapped
id pool image snap device
0 rbd test-image - /dev/rbd0
# 执行 rbd unmap
[root@k8s-node-42 ~]# rbd unmap /dev/rbd0
# 最后执行 rbd rm
[root@ceph-node-1 ~]# rbd rm test-image
Removing image: 100% complete...done.
|