在 KubeSphere 中使用 Rook 构建云原生存储环境
共 14574字,需浏览 30分钟
·
2022-01-10 18:00
Rook 介绍
Rook 是一个开源的云原生存储编排器,为各种存储解决方案提供平台、框架和支持,以便与云原生环境进行原生集成。
Rook 将分布式存储系统转变为自管理、自扩展、自修复的存储服务。它使存储管理员的部署、引导、配置、配置、扩展、升级、迁移、灾难恢复、监控和资源管理等任务自动化。
简而言之,Rook 就是一组 Kubernetes 的 Operator,它可以完全控制多种数据存储解决方案(例如 Ceph、EdgeFS、Minio、Cassandra)的部署,管理以及自动恢复。
到目前为止,Rook 支持的最稳定的存储仍然是 Ceph,本文将介绍如何使用 Rook 来创建维护 Ceph 集群,并作为 Kubernetes 的持久化存储。
环境准备
K8s 环境可以通过安装 KubeSphere 进行部署,我使用的是高可用方案。
在公有云上安装 KubeSphere 参考文档:多节点安装[1]
⚠️ 注意:kube-node(5,6,7)的节点上分别有两块数据盘。
kube-master1 Ready master 118d v1.17.9
kube-master2 Ready master 118d v1.17.9
kube-master3 Ready master 118d v1.17.9
kube-node1 Ready worker 118d v1.17.9
kube-node2 Ready worker 118d v1.17.9
kube-node3 Ready worker 111d v1.17.9
kube-node4 Ready worker 111d v1.17.9
kube-node5 Ready worker 11d v1.17.9
kube-node6 Ready worker 11d v1.17.9
kube-node7 Ready worker 11d v1.17.9
安装前请确保 node 节点都安装上了 lvm2,否则会报错。
部署安装 Rook、Ceph 集群
1.克隆 Rook 仓库到本地
$ git clone -b release-1.4 https://github.com/rook/rook.git
2.切换目录
$ cd /root/ceph/rook/cluster/examples/kubernetes/ceph
3.部署 Rook,创建 CRD 资源
$ kubectl create -f common.yaml -f operator.yaml
# 说明:
# 1.comm.yaml里面主要是权限控制以及CRD资源定义
# 2.operator.yaml是rook-ceph-operator的deloyment
4.创建 Ceph 集群
$ kubectl create -f cluster.yaml
# 重要说明:
# 演示不做定制化操作,Ceph集群默认会动态去识别node节点上未格式化的全新空闲硬盘,自动会对这些盘进行OSD初始化(至少是需要3个节点,每个节点至少一块空闲硬盘)
5.检查 pod 状态
$ kubectl get pod -n rook-ceph -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
csi-cephfsplugin-5fw92 3/3 Running 6 12d 192.168.0.31 kube-node7
csi-cephfsplugin-78plf 3/3 Running 0 12d 192.168.0.134 kube-node1
csi-cephfsplugin-bkdl8 3/3 Running 3 12d 192.168.0.195 kube-node5
csi-cephfsplugin-provisioner-77f457bcb9-6w4cv 6/6 Running 0 12d 10.233.77.95 kube-node4
csi-cephfsplugin-provisioner-77f457bcb9-q7vxh 6/6 Running 0 12d 10.233.76.156 kube-node3
csi-cephfsplugin-rqb4d 3/3 Running 0 12d 192.168.0.183 kube-node4
csi-cephfsplugin-vmrfj 3/3 Running 0 12d 192.168.0.91 kube-node3
csi-cephfsplugin-wglsw 3/3 Running 3 12d 192.168.0.116 kube-node6
csi-rbdplugin-4m8hv 3/3 Running 0 12d 192.168.0.91 kube-node3
csi-rbdplugin-7wt45 3/3 Running 3 12d 192.168.0.195 kube-node5
csi-rbdplugin-bn5pn 3/3 Running 3 12d 192.168.0.116 kube-node6
csi-rbdplugin-hwl4b 3/3 Running 6 12d 192.168.0.31 kube-node7
csi-rbdplugin-provisioner-7897f5855-7m95p 6/6 Running 0 12d 10.233.77.94 kube-node4
csi-rbdplugin-provisioner-7897f5855-btwt5 6/6 Running 0 12d 10.233.76.155 kube-node3
csi-rbdplugin-qvksp 3/3 Running 0 12d 192.168.0.183 kube-node4
csi-rbdplugin-rr296 3/3 Running 0 12d 192.168.0.134 kube-node1
rook-ceph-crashcollector-kube-node1-64cf6f49fb-bx8lz 1/1 Running 0 12d 10.233.101.46 kube-node1
rook-ceph-crashcollector-kube-node3-575b75dc64-gxwtp 1/1 Running 0 12d 10.233.76.149 kube-node3
rook-ceph-crashcollector-kube-node4-78549d6d7f-9zz5q 1/1 Running 0 8d 10.233.77.226 kube-node4
rook-ceph-crashcollector-kube-node5-5db8557476-b8zp6 1/1 Running 1 11d 10.233.81.239 kube-node5
rook-ceph-crashcollector-kube-node6-78b7946769-8qh45 1/1 Running 0 8d 10.233.66.252 kube-node6
rook-ceph-crashcollector-kube-node7-78c97898fd-k85l4 1/1 Running 1 8d 10.233.111.33 kube-node7
rook-ceph-mds-myfs-a-86bdb684b6-4pbj7 1/1 Running 0 8d 10.233.77.225 kube-node4
rook-ceph-mds-myfs-b-6697d66b7d-jgnkw 1/1 Running 0 8d 10.233.66.250 kube-node6
rook-ceph-mgr-a-658db99d5b-jbrzh 1/1 Running 0 12d 10.233.76.162 kube-node3
rook-ceph-mon-a-5cbf5947d8-vvfgf 1/1 Running 1 12d 10.233.101.44 kube-node1
rook-ceph-mon-b-6495c96d9d-b82st 1/1 Running 0 12d 10.233.76.144 kube-node3
rook-ceph-mon-d-dc4c6f4f9-rdfpg 1/1 Running 1 12d 10.233.66.219 kube-node6
rook-ceph-operator-56fc54bb77-9rswg 1/1 Running 0 12d 10.233.76.138 kube-node3
rook-ceph-osd-0-777979f6b4-jxtg9 1/1 Running 1 11d 10.233.81.237 kube-node5
rook-ceph-osd-10-589487764d-8bmpd 1/1 Running 0 8d 10.233.111.59 kube-node7
rook-ceph-osd-11-5b7dd4c7bc-m4nqz 1/1 Running 0 8d 10.233.111.60 kube-node7
rook-ceph-osd-2-54cbf4d9d8-qn4z7 1/1 Running 1 10d 10.233.66.222 kube-node6
rook-ceph-osd-6-c94cd566-ndgzd 1/1 Running 1 10d 10.233.81.238 kube-node5
rook-ceph-osd-7-d8cdc94fd-v2lm8 1/1 Running 0 9d 10.233.66.223 kube-node6
rook-ceph-osd-prepare-kube-node1-4bdch 0/1 Completed 0 66m 10.233.101.91 kube-node1
rook-ceph-osd-prepare-kube-node3-bg4wk 0/1 Completed 0 66m 10.233.76.252 kube-node3
rook-ceph-osd-prepare-kube-node4-r9dk4 0/1 Completed 0 66m 10.233.77.107 kube-node4
rook-ceph-osd-prepare-kube-node5-rbvcn 0/1 Completed 0 66m 10.233.81.73 kube-node5
rook-ceph-osd-prepare-kube-node5-rcngg 0/1 Completed 5 10d 10.233.81.98 kube-node5
rook-ceph-osd-prepare-kube-node6-jc8cm 0/1 Completed 0 66m 10.233.66.109 kube-node6
rook-ceph-osd-prepare-kube-node6-qsxrp 0/1 Completed 0 11d 10.233.66.109 kube-node6
rook-ceph-osd-prepare-kube-node7-5c52p 0/1 Completed 5 8d 10.233.111.58 kube-node7
rook-ceph-osd-prepare-kube-node7-h5d6c 0/1 Completed 0 66m 10.233.111.110 kube-node7
rook-ceph-osd-prepare-kube-node7-tzvp5 0/1 Completed 0 11d 10.233.111.102 kube-node7
rook-ceph-osd-prepare-kube-node7-wd6dt 0/1 Completed 7 8d 10.233.111.56 kube-node7
rook-ceph-tools-64fc489556-5clvj 1/1 Running 0 12d 10.233.77.118 kube-node4
rook-discover-6kbvg 1/1 Running 0 12d 10.233.101.42 kube-node1
rook-discover-7dr44 1/1 Running 2 12d 10.233.66.220 kube-node6
rook-discover-dqr82 1/1 Running 0 12d 10.233.77.74 kube-node4
rook-discover-gqppp 1/1 Running 0 12d 10.233.76.139 kube-node3
rook-discover-hdkxf 1/1 Running 1 12d 10.233.81.236 kube-node5
rook-discover-pzhsw 1/1 Running 3 12d 10.233.111.36 kube-node7
以上是所有组件的 pod 完成后的状态,其中 rook-ceph-osd-prepare 开头的 pod 是自动感知集群新挂载硬盘的,只要有新硬盘挂载到集群自动会触发 OSD。
6.配置 Ceph 集群 dashboard
Ceph Dashboard 是一个内置的基于 Web 的管理和监视应用程序,它是开源 Ceph 发行版的一部分。通过 Dashboard 可以获取 Ceph 集群的各种基本状态信息。
默认的 ceph 已经安装的 ceph-dashboard,其 SVC 地址是 service clusterIP,并不能被外部访问,需要创建 service 服务
$ kubectl apply -f dashboard-external-http.yaml
apiVersion: v1
kind: Service
metadata:
name: rook-ceph-mgr-dashboard-external-https
namespace: rook-ceph # namespace:cluster
labels:
app: rook-ceph-mgr
rook_cluster: rook-ceph # namespace:cluster
spec:
ports:
- name: dashboard
port: 7000
protocol: TCP
targetPort: 7000
selector:
app: rook-ceph-mgr
rook_cluster: rook-ceph
sessionAffinity: None
type: NodePort
说明:由于 8443 是 https 访问端口需要配置证书,本教程只展示 http 访问 port 上只配置了 7000
7.查看 svc 状态
$ kubectl get svc -n rook-ceph
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
csi-cephfsplugin-metrics ClusterIP 10.233.3.172 8080/TCP,8081/TCP 12d
csi-rbdplugin-metrics ClusterIP 10.233.43.23 8080/TCP,8081/TCP 12d
rook-ceph-mgr ClusterIP 10.233.63.85 9283/TCP 12d
rook-ceph-mgr-dashboard ClusterIP 10.233.20.159 7000/TCP 12d
rook-ceph-mgr-dashboard-external-https NodePort 10.233.56.73 7000:31357/TCP 12d
rook-ceph-mon-a ClusterIP 10.233.30.222 6789/TCP,3300/TCP 12d
rook-ceph-mon-b ClusterIP 10.233.55.25 6789/TCP,3300/TCP 12d
rook-ceph-mon-d ClusterIP 10.233.0.206 6789/TCP,3300/TCP 12d
8.验证访问 dashboard
打开 KubeSphere 平台开启外网服务
访问方式:
http://{master1-ip:31357}
用户名获取方法:
$ kubectl -n rook-ceph get secret rook-ceph-dashboard-password -o jsonpath="{['data']['password']}"|base64 --decode && echo
说明:dashboard 显示 HEALTH_WARN 警告可以通过 seelog 的方式查看具体的原因,一般是 osd down、pg 数量不够等
9.部署 rook 工具箱
Rook 工具箱是一个包含用于 Rook 调试和测试的常用工具的容器
$ kubectl apply -f toolbox.yaml
进入工具箱查看 Ceph 集群状态
$ kubectl -n rook-ceph exec -it $(kubectl -n rook-ceph get pod -l "app=rook-ceph-tools" -o jsonpath='{.items[0].metadata.name}') -- bash
$ ceph -s
cluster:
id: 1457045a-4926-411f-8be8-c7a958351a38
health: HEALTH_WARN
mon a is low on available space
2 osds down
Degraded data redundancy: 25/159 objects degraded (15.723%), 16 pgs degraded, 51 pgs undersized
3 daemons have recently crashed
services:
mon: 3 daemons, quorum a,b,d (age 9d)
mgr: a(active, since 4h)
mds: myfs:1 {0=myfs-b=up:active} 1 up:standby-replay
osd: 12 osds: 6 up (since 8d), 8 in (since 8d); 9 remapped pgs
data:
pools: 5 pools, 129 pgs
objects: 53 objects, 37 MiB
usage: 6.8 GiB used, 293 GiB / 300 GiB avail
pgs: 25/159 objects degraded (15.723%)
5/159 objects misplaced (3.145%)
69 active+clean
35 active+undersized
16 active+undersized+degraded
9 active+clean+remapped
工具箱相关查询命令
ceph status
ceph osd status
ceph df
rados df
部署 StorageClass
1.rbd 块存储简介
Ceph 可以同时提供对象存储 RADOSGW、块存储 RBD、文件系统存储 Ceph FS。RBD 即 RADOS Block Device 的简称,RBD 块存储是最稳定且最常用的存储类型。RBD 块设备类似磁盘可以被挂载。RBD 块设备具有快照、多副本、克隆和一致性等特性,数据以条带化的方式存储在 Ceph 集群的多个 OSD 中。
2.创建 StorageClass
[root@kube-master1 rbd]# kubectl apply -f storageclass.yaml
3.查看 StorageClass 部署状态
4.创建 pvc
$ kubectl apply -f pvc.yaml
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: rbd-pvc
spec:
accessModes:
- ReadWriteOnce
resources:
requests:
storage: 2Gi
storageClassName: rook-ceph-block
~
5.创建带有 pvc 的 pod
$ kubectl apply -f pod.yaml
apiVersion: v1
kind: Pod
metadata:
name: csirbd-demo-pod
spec:
containers:
- name: web-server
image: nginx
volumeMounts:
- name: mypvc
mountPath: /var/lib/www/html
volumes:
- name: mypvc
persistentVolumeClaim:
claimName: rbd-pvc
readOnly: false
6.查看 pod、pvc、pv 状态
总结
对于首次接触 rook+Ceph 部署体验的同学来说需要了解的内容较多,遇到的坑也会比较的多。希望通过以上的部署过程记录可以帮助到大家。
1.Ceph 集群一直提示没有可 osd 的盘
答:这里遇到过几个情况,查看下挂载的数据盘是不是以前已经使用过虽然格式化了但是以前的 raid 信息还存在?可以使用一下脚本进行清理后在格式化在进行挂载。
#!/usr/bin/env bash
DISK="/dev/vdc" #按需修改自己的盘符信息
# Zap the disk to a fresh, usable state (zap-all is important, b/c MBR has to be clean)
# You will have to run this step for all disks.
sgdisk --zap-all $DISK
# Clean hdds with dd
dd if=/dev/zero of="$DISK" bs=1M count=100 oflag=direct,dsync
# Clean disks such as ssd with blkdiscard instead of dd
blkdiscard $DISK
# These steps only have to be run once on each node
# If rook sets up osds using ceph-volume, teardown leaves some devices mapped that lock the disks.
ls /dev/mapper/ceph-* | xargs -I% -- dmsetup remove %
# ceph-volume setup can leave ceph- directories in /dev and /dev/mapper (unnecessary clutter)
rm -rf /dev/ceph-*
rm -rf /dev/mapper/ceph--*
# Inform the OS of partition table changes
partprobe $DISK
~
2.Ceph 支持哪些存储类型?
答:rdb 块存储、cephfs 文件存储、s3 对象存储等
3.部署中出现各种坑应该怎么排查?
答:强烈建议通过 rook、ceph 官网去查看相关文档进行排错
https://rook.github.io/docs/rook/ https://docs.ceph.com/en/pacific/
4.访问 dashboard 失败
答:如果是公有云搭建的 KubeSphere 或 K8s 请把 nodeport 端口在安全组里放行即可
引用链接
多节点安装: https://v3-1.docs.kubesphere.io/zh/docs/installing-on-linux/public-cloud/install-kubesphere-on-huaweicloud-ecs/
关于 KubeSphere
KubeSphere (https://kubesphere.io)是在 Kubernetes 之上构建的开源容器混合云,提供全栈的 IT 自动化运维的能力,简化企业的 DevOps 工作流。
KubeSphere 已被 Aqara 智能家居、爱立信、本来生活、东软、华云、新浪、三一重工、华夏银行、四川航空、国药集团、微众银行、杭州数跑科技、紫金保险、去哪儿网、中通、中国人民银行、中国银行、中国人保寿险、中国太平保险、中国移动、中国电信、天翼云、中移金科、Radore、ZaloPay 等海内外数千家企业采用。KubeSphere 提供了开发者友好的向导式操作界面和丰富的企业级功能,包括 Kubernetes 多云与多集群管理、DevOps (CI/CD)、应用生命周期管理、边缘计算、微服务治理 (Service Mesh)、多租户管理、可观测性、存储与网络管理、GPU support 等功能,帮助企业快速构建一个强大和功能丰富的容器云平台。