基于kubesphere的k8s环境部署三节点的rook-ceph集群-Toy模板网

这篇具有很好参考价值的文章主要介绍了基于kubesphere的k8s环境部署三节点的rook-ceph集群。希望对大家有所帮助。如果存在错误或未考虑完全的地方，请大家不吝赐教，您也可以点击"举报违法"按钮提交疑问。

前言

这次实验记录的是使用虚拟机搭建的使用了三个计算\存储节点的rook-ceph集群，模拟真实的使用场景。补充之前发的只有单点的部署方式。

链接：基于kubesphere的k8s环境部署单点版本的rook-ceph

一、rook-ceph是什么？

Rook is an open source cloud-native storage orchestrator, providing the platform, framework, and support for Ceph storage to natively integrate with cloud-native environments.
Rook是一个开源的云原生存储编排器，为Ceph存储提供平台、框架和支持，以便与云原生环境进行本地集成。

二、开始部署

2.1 环境准备

虚拟机数量：四台
虚拟机镜像类型：CentOS-7-x86_64-Minimal-2009.iso
k8s环境：一套
k8s环境:1.23.6

机器列表如下

hostname	IP	系统盘	数据盘
kubeadmin	192.168.150.61	sda(50G)	无
kubeworker01	192.168.150.62	sda(20G)	vda(20G),vdb(20G)
kubeworker02	192.168.150.63	sda(20G)	vda(20G),vdb(20G)
kubeworker03	192.168.150.64	sda(20G)	vda(20G),vdb(20G)

2.2 软件包准备，计算\存储节点执行

安装软件包，加载rbd模块

#软件包装备
yum install -y git lvm2 gdisk
#内核加载rbd模块
modprobe rbd
lsmod | grep rbd

备注：删除残留数据，如果部署失败，一定清理下数据，不清理的话会影响下一次的部署

删除配置文件目录
rm -rf /var/lib/rook/
格式化磁盘
gdisk --zap-all /dev/vda
gdisk --zap-all /dev/vdb
dd if=/dev/zero of=/dev/vda  bs=1M count=100 oflag=direct,dsync
dd if=/dev/zero of=/dev/vdb  bs=1M count=100 oflag=direct,dsync

2.3 下载rook-ceph文件

下载文件并提取核心文件到自己的部署文件夹

cd /data/
yum install -y git
git clone --single-branch --branch v1.11.6 https://github.com/rook/rook.git
# 提取部署文件
mkdir -p /data/rook-ceph/
cp /data/rook/deploy/examples/crds.yaml /data/rook-ceph/crds.yaml
cp /data/rook/deploy/examples/common.yaml /data/rook-ceph/common.yaml
cp /data/rook/deploy/examples/operator.yaml /data/rook-ceph/operator.yaml
cp /data/rook/deploy/examples/cluster.yaml /data/rook-ceph/cluster.yaml
cp /data/rook/deploy/examples/filesystem.yaml /data/rook-ceph/filesystem.yaml
cp /data/rook/deploy/examples/toolbox.yaml /data/rook-ceph/toolbox.yaml
cp /data/rook/deploy/examples/csi/rbd/storageclass.yaml /data/rook-ceph/storageclass-rbd.yaml
cp /data/rook/deploy/examples/csi/cephfs/storageclass.yaml /data/rook-ceph/storageclass-cephfs.yaml
cp /data/rook/deploy/examples/csi/nfs/storageclass.yaml /data/rook-ceph/storageclass-nfs.yaml

cd /data/rook-ceph

2.4 部署operator

修改镜像仓库信息，operator.yaml中镜像仓库修改为阿里云的镜像仓库配置

ROOK_CSI_CEPH_IMAGE: "quay.io/cephcsi/cephcsi:v3.8.0"
ROOK_CSI_REGISTRAR_IMAGE: "registry.cn-hangzhou.aliyuncs.com/google_containers/csi-node-driver-registrar:v2.7.0"
ROOK_CSI_RESIZER_IMAGE: "registry.cn-hangzhou.aliyuncs.com/google_containers/csi-resizer:v1.7.0"
ROOK_CSI_PROVISIONER_IMAGE: "registry.cn-hangzhou.aliyuncs.com/google_containers/csi-provisioner:v3.4.0"
ROOK_CSI_SNAPSHOTTER_IMAGE: "registry.cn-hangzhou.aliyuncs.com/google_containers/csi-snapshotter:v6.2.1"
ROOK_CSI_ATTACHER_IMAGE: "registry.cn-hangzhou.aliyuncs.com/google_containers/csi-attacher:v4.1.0"

执行部署

# 开始部署
cd /data/rook-ceph
kubectl create -f crds.yaml
kubectl create -f common.yaml
kubectl create -f operator.yaml
# 检查operator的创建运行状态
kubectl -n rook-ceph get pod
# 输出
NAME                                 READY   STATUS    RESTARTS   AGE
rook-ceph-operator-585f6875d-qjhdn   1/1     Running   0          4m36s

2.5 创建ceph集群

修改cluster.yaml，配置osd对应的数据盘,这里截取修改的部分，主要是nodes这里的配置。我这里按照集群的现有硬盘，将三个节点的数据盘都直接配置上了。没有使用规则发现或是全部发现。

  priorityClassNames:
    mon: system-node-critical
    osd: system-node-critical
    mgr: system-cluster-critical
    useAllNodes: false
    useAllDevices: false
    deviceFilter: 
    config:
    nodes:
      - name: "kubeworker01"
        devices: 
          - name: "vda" 
          - name: "vdb" 
      - name: "kubeworker02"
        devices:
          - name: "vda" 
          - name: "vdb" 
      - name: "kubeworker03"
        devices:
          - name: "vda"           
          - name: "vdb"

执行部署cluster-test.yaml

kubectl create -f cluster.yaml
# 会部署一段时间
kubectl -n rook-ceph get pod
# 查看部署结果，当全部为Running之后部署工具容器进行集群确认
NAME                                                     READY   STATUS      RESTARTS   AGE
csi-cephfsplugin-7qk26                                   2/2     Running     0          34m
csi-cephfsplugin-dp8zx                                   2/2     Running     0          34m
csi-cephfsplugin-fb6rh                                   2/2     Running     0          34m
csi-cephfsplugin-provisioner-5549b4bcff-56ntx            5/5     Running     0          34m
csi-cephfsplugin-provisioner-5549b4bcff-m5j76            5/5     Running     0          34m
csi-rbdplugin-d829n                                      2/2     Running     0          34m
csi-rbdplugin-provisioner-bcff85bf9-7thl7                5/5     Running     0          34m
csi-rbdplugin-provisioner-bcff85bf9-cctkc                5/5     Running     0          34m
csi-rbdplugin-rj9wp                                      2/2     Running     0          34m
csi-rbdplugin-zs6s2                                      2/2     Running     0          34m
rook-ceph-crashcollector-kubeworker01-794647548b-bdrcx   1/1     Running     0          91s
rook-ceph-crashcollector-kubeworker02-d97cfb685-ss2sl    1/1     Running     0          86s
rook-ceph-crashcollector-kubeworker03-9d65c8dd8-zrv5x    1/1     Running     0          22m
rook-ceph-mgr-a-6fccb8744f-5zdvf                         3/3     Running     0          23m
rook-ceph-mgr-b-7c4bbbfcf4-fhxm9                         3/3     Running     0          23m
rook-ceph-mon-a-56dc4dfb8d-4j2bz                         2/2     Running     0          34m
rook-ceph-mon-b-7d6d96649b-spz4p                         2/2     Running     0          33m
rook-ceph-mon-c-759c774dc7-8hftq                         2/2     Running     0          28m
rook-ceph-operator-f45db9b9f-knbx4                       1/1     Running     0          2m9s
rook-ceph-osd-0-86cd7776c8-bm764                         2/2     Running     0          91s
rook-ceph-osd-1-7686cf9757-ss9z2                         2/2     Running     0          86s
rook-ceph-osd-2-5bc55847d-g2z6l                          2/2     Running     0          91s
rook-ceph-osd-3-998bccb64-rq9cf                          2/2     Running     0          83s
rook-ceph-osd-4-5c7c7f555b-djdvl                         2/2     Running     0          86s
rook-ceph-osd-5-69976f85fc-9xz94                         2/2     Running     0          83s
rook-ceph-osd-prepare-kubeworker01-qlvcp                 0/1     Completed   0          104s
rook-ceph-osd-prepare-kubeworker02-mnhcj                 0/1     Completed   0          100s
rook-ceph-osd-prepare-kubeworker03-sbk76                 0/1     Completed   0          97s
rook-ceph-tools-598b59df89-77sm7                         1/1     Running     0          7m43s

2.6 创建工具容器，检查集群状态

# 创建工具容器
kubectl apply -f toolbox.yaml
# 执行命令查询
kubectl -n rook-ceph exec -it deploy/rook-ceph-tools -- ceph -s
# 输出：
  cluster:
    id:     3a04d434-a2ac-4f2a-a231-a08ca46c6df3
    health: HEALTH_OK
 
  services:
    mon: 3 daemons, quorum a,b,c (age 20m)
    mgr: a(active, since 4m), standbys: b
    osd: 6 osds: 6 up (since 3m), 6 in (since 4m)
 
  data:
    pools:   1 pools, 1 pgs
    objects: 2 objects, 449 KiB
    usage:   521 MiB used, 119 GiB / 120 GiB avail
    pgs:     1 active+clean

# 或是进入工具容器内执行命令
kubectl -n rook-ceph exec -it deploy/rook-ceph-tools -- bash
# 查看集群状态
bash-4.4$ ceph -s 
  cluster:
    id:     3a04d434-a2ac-4f2a-a231-a08ca46c6df3
    health: HEALTH_OK
 
  services:
    mon: 3 daemons, quorum a,b,c (age 21m)
    mgr: a(active, since 5m), standbys: b
    osd: 6 osds: 6 up (since 5m), 6 in (since 5m)
 
  data:
    pools:   1 pools, 1 pgs
    objects: 2 objects, 449 KiB
    usage:   521 MiB used, 119 GiB / 120 GiB avail
    pgs:     1 active+clean

2.7 准备dashboard的nodeport端口映射服务

cat > /data/rook-ceph/dashboard-external-https.yaml <<EOF
apiVersion: v1
kind: Service
metadata:
  name: rook-ceph-mgr-dashboard-external-https
  namespace: rook-ceph
  labels:
    app: rook-ceph-mgr
    rook_cluster: rook-ceph
spec:
  ports:
  - name: dashboard
    port: 8443
    protocol: TCP
    targetPort: 8443
    nodePort: 30808
  selector:
    app: rook-ceph-mgr
    rook_cluster: rook-ceph
  sessionAffinity: None
  type: NodePort

EOF

# 这里的nodeport端口建议更换为适合自己环境规划的端口
kubectl apply -f dashboard-external-https.yaml
# 输出
service/rook-ceph-mgr-dashboard-external-https created
# 获取admin用户密码
kubectl -n rook-ceph get secret rook-ceph-dashboard-password -o jsonpath="{['data']['password']}" | base64 --decode && echo

使用浏览器访问端口192.168.150.61:30808，使用admin用户登陆，登陆后可以修改密码，也可以新建用户
基于kubesphere的k8s环境部署三节点的rook-ceph集群

成功登陆
基于kubesphere的k8s环境部署三节点的rook-ceph集群

2.8 准备prometheus的metric端口映射服务

cat > /data/rook-ceph/metric-external-https.yaml <<EOF
apiVersion: v1
kind: Service
metadata:
  name: rook-ceph-mgr-metric-external-https
  namespace: rook-ceph
  labels:
    app: rook-ceph-mgr
    rook_cluster: rook-ceph
spec:
  ports:
  - name: metric
    port: 9283
    protocol: TCP
    targetPort: 9283
    nodePort: 30809
  selector:
    app: rook-ceph-mgr
    rook_cluster: rook-ceph
  sessionAffinity: None
  type: NodePort
EOF

# 这里的nodeport端口建议更换为适合自己环境规划的端口
kubectl apply -f metric-external-https.yaml
# 输出
service/rook-ceph-mgr-metric-external-https created

使用浏览器访问端口192.168.150.61:30809
基于kubesphere的k8s环境部署三节点的rook-ceph集群

三、创建存储类

3.1 创建cephrbd存储类

kubectl apply -f  storageclass-rbd.yaml

创建测试用的pvc

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: test-rbd-pv-claim
spec:
  storageClassName: rook-ceph-block
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 10G

3.2 创建cephfs存储类

kubectl apply -f filesystem.yaml
kubectl apply  -f storageclass-cephfs.yaml
# 创建filesystem.yaml之后会生成rook-ceph-mds-myfs的工作负载
kubectl -n rook-ceph get pod |grep mds
# 输出
rook-ceph-mds-myfs-a-5d5754b77-nlcb9                     2/2     Running     0               97s
rook-ceph-mds-myfs-b-9f9dd7f6-sc6qm                      2/2     Running     0               96s

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: test-cephfs-pv-claim
spec:
  storageClassName: rook-cephfs
  accessModes:
    - ReadWriteMany
  resources:
    requests:
      storage: 10G

3.3 查看创建结果

查询存储

kubectl get storageclass
# 输出
NAME              PROVISIONER                     RECLAIMPOLICY   VOLUMEBINDINGMODE      ALLOWVOLUMEEXPANSION   AGE
local (default)   openebs.io/local                Delete          WaitForFirstConsumer   false                  21d
rook-ceph-block   rook-ceph.rbd.csi.ceph.com      Delete          Immediate              true                   4s
rook-cephfs       rook-ceph.cephfs.csi.ceph.com   Delete          Immediate              true                   4m44s

查询pvc

kubectl  get pvc -o wide
# 输出
NAME                   STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS      AGE   VOLUMEMODE
test-cephfs-pv-claim   Bound    pvc-3cdd9e88-2ae2-4e23-9f23-13e095707964   10Gi       RWX            rook-cephfs       7s    Filesystem
test-rbd-pv-claim      Bound    pvc-55a57b74-b595-4726-8b82-5257fd2d279a   10Gi       RWO            rook-ceph-block   6s    Filesystem

补充

调试方式

当集群没有按照预期成功部署的情况下，可以执行如下的命令重新部署或是查看异常的原因

# 重启ceph operator调度，重新部署
kubectl rollout restart deploy rook-ceph-operator -n rook-ceph
#注：如果新osd pod无法执行起来可以通过查询osd prepare日志找问题
kubectl -n rook-ceph logs rook-ceph-osd-prepare-nodeX-XXXXX provision
#查看状态
kubectl -n rook-ceph get CephCluster -o yaml

关于使用裸分区的实验，不推荐生产使用

用虚拟机测试了一下裸分区作为osd的存储介质，也可以运行。但是压力上来之后，对其余的分区存在较大影响。因为我实验的时候是把系统盘的一个大分区用作存储介质了，导致系统分区压力升高。etcd响应时间出现报错，k8s相关服务出现异常。不推荐生产使用。

  priorityClassNames:
    mon: system-node-critical
    osd: system-node-critical
    mgr: system-cluster-critical
    useAllNodes: false
    useAllDevices: false
    deviceFilter: 
    config:
    nodes:
    - name: "kubeworker01"
      devices: 
        - name: "vda6" 
    - name: "kubeworker02"
      devices:
        - name: "vda6"  
    - name: "kubeworker03"
      devices:
        - name: "vda6"