今天尝试学习一下k8s的pv&pvc,并记录下实验的过程
pv&pvc简介
k8s提供了emptyDir,hostPath,rbd,cephfs等存储方式供容器使用,不过这些存储方式都有一个缺点:开发人员必须得知指定存储的相关配置信息,才能使用存储.例如要使用cephfs,Pod的配置信息就必须指明cephfs的monitor,user,selectFile等等,而这些应该是系统管理员的工作.对此,k8s提供了两个新的API资源:PersistentVolume,PersistentVolumeClaim
PV(PersistentVolume)是管理员已经提供好的一块存储.在k8s集群中,PV像Node一样,是一个资源
PVC(PersistentVolumeClaim)是用户对PV的一次申请.PVC对于PV就像Pod对于Node一样,Pod可以申请CPU和Memory资源,而PVC也可以申请PV的大小与权限
有了PersistentVolumeClaim,用户只需要告诉Kubernetes需要什么样的存储资源,而不必关心真正的空间从哪里分配,如何访问等底层细节信息;这些Storage Provider的底层信息交给管理员来处理,只有管理员才应该关心创建PersistentVolume的细节信息
实验
实验环境:
192.168.122.10:host1,Master
192.168.122.20:host2,Node
192.168.122.30:host3,Node
目的:使用nfs类型的PV,并使用Job验证PV创建是否创建成功
1. 在host1上创建nfs服务器,用于提供存储
安装nfs服务端:1
2
3
4[[email protected] kube]# yum install nfs-utils rpcbind
...
Installed:
nfs-utils.x86_64 1:1.3.0-0.54.el7 rpcbind.x86_64 0:0.2.0-44.el7
创建用于nfs服务的存储文件夹:1[[email protected] kube]# mkdir -p /data/nfs
编辑/etc/exports文件:1
2[[email protected] kube]# vim /etc/exports
/data/nfs 192.168.122.0/24(rw,sync)
开启nfs服务:1[[email protected] kube]# systemctl start nfs-server.service
检验是否开启成功:1
2
3[[email protected] kube]# showmount -e
Export list for host1:
/data/nfs 192.168.122.0/24
2. 创建pv
编辑pv资源的配置文件:1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21~]$ cat pv_nfs.yml
apiVersion: v1
kind: PersistentVolume
metadata:
name: nfspv1
spec:
capacity:
storage: 1Gi
#指定访问模式
accessModes:
#pv能以readwrite模式mount到单个节点
- ReadWriteOnce
#指定pv的回收策略,即pvc资源释放后的事件.recycle(不建议,使用动态供给代替)删除pvc的所有文件
persistentVolumeReclaimPolicy: Recycle
#指定pv的class为nfs,相当于为pv分类,pvc将指定class申请pv
storageClassName: mynfs
#指定pv为nfs服务器上对应的目录
nfs:
path: /data/nfs
server: 192.168.122.10
应用该pv资源:1
2[[email protected] ~]$ kubectl apply -f pv_nfs.yml
persistentvolume/nfspv1 created
查看该pv资源:1
2
3[[email protected] ~]$ kubectl get pv
NAME CAPACITY ACCESS MODES RECLAIM POLICY STATUS CLAIM STORAGECLASS REASON AGE
nfspv1 1Gi RWO Recycle Available mynfs 14s
3. 创建pvc
编辑pvc资源配置文件:1
2
3
4
5
6
7
8
9
10
11
12~]$ cat pvc_nfs.yml
kind: PersistentVolumeClaim
apiVersion: v1
metadata:
name: nfspvc1
spec:
accessModes:
- ReadWriteOnce
resources:
requests:
storage: 1Gi
storageClassName: mynfs
应用pvc配置文件:1
2[[email protected] ~]$ kubectl apply -f pvc_nfs.yml
persistentvolumeclaim/nfspvc1 created
查看该pvc资源:1
2
3[[email protected] ~]$ kubectl get pvc
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
nfspvc1 Bound nfspv1 1Gi RWO mynfs 24s
4. 创建应用资源,使用pvc存储
编辑应用资源配置文件:1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23apiVersion: batch/v1
kind: Job
metadata:
name: pvjob
spec:
template:
spec:
containers:
- name: bbox1
image: busybox
args:
- /bin/sh
- -c
- echo "hello pv" > /mydata/hello
volumeMounts:
- mountPath: "/mydata"
name: mydata
restartPolicy: Never
volumes:
- name: mydata
persistentVolumeClaim:
claimName: nfspvc1
该job将在nfs的volume创建一个hello文件,打印”hello pv”字符串
应用该Job资源:1
2[[email protected] ~]$ kubectl apply -f pvjob.yml
job.batch/pvjob created
查看host1上的/data/nfs目录,检查实验结果:1
2[[email protected] ~]$ cat /data/nfs/hello
hello pv
实验成功!
排错
在这次实验中出现了几个错误:
1. 使用pvc存储的pod一直处于pending状态,无法启动1
2
3
4
5
6[[email protected] ~]$ kubectl get pod
NAME READY STATUS RESTARTS AGE
pvjob-47cwq 0/1 Pending 0 5s
[[email protected] ~]$ kubectl get pod
NAME READY STATUS RESTARTS AGE
pvjob-47cwq 0/1 Pending 0 3m
查看该pod的启动事件:1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21[[email protected] ~]$ kubectl describe pod pvjob-47cwq
Name: pvjob-47cwq
Namespace: default
Node: host3/192.168.122.30
Status: Pending
IP:
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Scheduled 44s default-scheduler Successfully assigned default/liveness to host3
Warning FailedMount 44s kubelet, host3 MountVolume.SetUp failed for volume "nfspv1" : mount failed: exit status 32
Mounting command: systemd-run
Mounting arguments: --description=Kubernetes transient mount for /var/lib/kubelet/pods/91ddd2ee-a2b8-11e8-a8f5-525400909158/volumes/kubernetes.io~nfs/nfspv1 --scope -- mount -t nfs 192.168.122.10:/data/nfs /var/lib/kubelet/pods/91ddd2ee-a2b8-11e8-a8f5-525400909158/volumes/kubernetes.io~nfs/nfspv1
Output: Running scope as unit run-24516.scope.
mount: wrong fs type, bad option, bad superblock on 192.168.122.10:/data/nfs,
missing codepage or helper program, or other error
(for several filesystems (e.g. nfs, cifs) you might
need a /sbin/mount. helper program)
In some cases useful info is found in syslog - try
dmesg | tail or so.
注意到报错信息:wrong fs type, bad option, bad superblock on 192.168.122.10:/data/nfs,这是因为host3没有安装nfs-utils软件包,无法识别nfs类型的文件系统,也无法作为nfs的客户端使用
解决方案:安装nfs-utils软件包,删除之前创建失败的Job资源并重新创建1
2
3
4
5
6
7
8
9
10[[email protected] ~]# yum install nfs-utils -y
...
Installed:
nfs-utils.x86_64 1:1.3.0-0.54.el7
[[email protected] ~]$ kubectl delete pod pvjob-47cwq
pod "pvjob-47cwq" deleted
[[email protected] ~]$ kubectl apply -f pvjob-47cwq
job.batch/pvjob created
可以看到,Job资源正常创建:1
2
3[[email protected] ~]$ kubectl get pod
NAME READY STATUS RESTARTS AGE
pvjob-47cwq 0/1 ContainerCreating 0 6s
2. 使用pvc的job执行失败1
2
3
4
5[[email protected] ~]$ kubectl get pod
NAME READY STATUS RESTARTS AGE
pvjob-c25xg 0/1 Error 0 25s
pvjob-db6fb 0/1 Error 0 32s
pvjob-hf7px 0/1 Error 0 15s
查看执行失败Job的log日志:1
2[[email protected] ~]$ kubectl logs pvjob-74g64
/bin/sh: can't create /mydata/hello: Permission denied
解决方案:
Permission denied多见于普通用户执行高权限命令失败,不过busybox容器本身使用的就是root用户,因此不存在这个问题.在nfs中,nfs服务端没有权限访问挂载的目录也会导致这个问题
更改目录属主为nfsnobody:1[[email protected] ~]$ sudo chown nfsnobody /data/nfs
删除原来Job并重新创建:1
2
3
4
5[[email protected] ~]$ kubectl delete jobs pvjob.yml
job.batch "pvjob" deleted
[[email protected] ~]$ kubectl apply -f pvjob.yml
job.batch/pvjob created
可以看到,Job执行成功:1
2
3[[email protected] ~]$ kubectl get pod
NAME READY STATUS RESTARTS AGE
pvjob-pct6b 0/1 Completed 0 8s