k8s集群使用cinder作为后端存储

说明

官方的部署文档中说明,使用cinder csi需要OpenStack作为Kubernetes Provider(这点还是很限制cinder在k8s中的使用)。因为对OpenStack不了解(也对“需要OpenStack作为Kubernetes Provider”不太理解),我直接在自己的k8s测试集群中部署cinder csi,csi驱动的容器都能正常启,创建pvc也可以动态绑定pv(cinder 的volume可以创建成功),但是创建pod时,csi controller attach这个volume到node上时失败了。原因是:自己本地搭建k8s的话,它申请的vloume资源 openstack没办法挂载到k8s服务器。下面记录下OpenStack作为Provider的k8s如何使用cinder。

环境

可以按官方文档搭建OpenStack和k8s集群。我用OpenStack是虚拟化团队创建的。k8s版本v1.18.3,CSI Cinder driver release.1.17

部署cinder csi

  1. clonecloud-provider-openstack项目
[root@kubernetes-master home]# git clone -b release-1.17 https://github.com/kubernetes/cloud-provider-openstack.git
[root@kubernetes-master home]# cd cloud-provider-openstack/manifests/cinder-csi-plugin
[root@kubernetes-master cinder-csi-plugin]# ls -lrt
total 28
-rw-r--r--. 1 root root  199 Jul 14 10:11 csi-cinder-driver.yaml
-rw-r--r--. 1 root root 3467 Jul 14 10:11 cinder-csi-nodeplugin.yaml
-rw-r--r--. 1 root root  694 Jul 14 10:11 cinder-csi-nodeplugin-rbac.yaml
-rw-r--r--. 1 root root 3545 Jul 14 10:11 cinder-csi-controllerplugin.yaml
-rw-r--r--. 1 root root 5555 Jul 14 10:11 cinder-csi-controllerplugin-rbac.yaml
-rw-r--r--. 1 root root  800 Jul 14 10:20 csi-secret-cinderplugin.yaml
  1. 根据openstack接口信息配置cloud.conf, 放到cat /etc/kubernetes/cloud.conf
[root@kubernetes-master cinder-csi-plugin]# cat /etc/kubernetes/cloud.conf
[Global]
username = admin
password = ZE0LNthHkgfS7M13KjA6aBdZ4A1x3w3nd
domain-name = default
auth-url = http://10.145.69.61:35357/v3
tenant-id = 6441367ec4ae4ce39e5f265dd
region = bjcc_region_2
  1. 修改csi-secret-cinderplugin.yam
  • 在OpenStack中,compute实例使用配置驱动器或元数据服务来检索特定于实例的数据,所以按照官方文档我们将[Metadata]节加到cloud.conf(为了给csi驱动使用,所以不需要更改我们之前创建的/etc/kubernetes/cloud.conf)
[root@kubernetes-master home]# cat cloud.conf 
[Global]
username = admin
password = ZE0LNthHkgfS7M13KjA6aBdZ4A1x3w3nd
domain-name = default
auth-url = http://10.145.69.61:35357/v3
tenant-id = 6441367ec4ae4ce39e5f265dd
region = bjcc_region_2

[Metadata]
search-order = configDrive,metadataService
  • 对cloud.conf做base64
[root@kubernetes-master home]# cat cloud.conf | base64 -w 0 
W0dsb2JhbF0KdXNlcm5hbWUgPSBhZG1pbgpwYXNzd29yZCA9IFpFMExOdGhIa2dmUzdNMTNLakE2YUJkWjRBMXgzdzNuZE5VS2x1WU8KZG9tYWluLW5hbWUgPSBkZWZhdWx0CmF1dGgtdXJsID0gaHR0cDovLzEwLjE0NS42OS42OjM1MzU3L3YzCnRlbmFudC1pZCA9IDY0NDEzNjdlYzRhZTRjZTM5ZTVmMjY1ZGQwMWRiY2IxCnJlZ2lvbiA9IGJqY2NfcmVnaW9uXzIKCltNZXRhZGF0YV0Kc2VhcmNoLW9yZGVyID0gY29uZmlnRHJpdmUsbWV0YWRhdGFTZXJ2aWNlCg==
  • 修改csi-secret-cinderplugin.yaml
[root@kubernetes-master cinder-csi-plugin]# cat csi-secret-cinderplugin.yaml 
# This YAML file contains secret objects,
# which are necessary to run csi cinder plugin.

kind: Secret
apiVersion: v1
metadata:
  name: cloud-config
  namespace: kube-system
data:
  #cloud.conf: W0dsb2JhbF0KdXNlcm5hbWUgPSBhZG1pbgpwYXNzd29yZCA9IG5vbW9yZXNlY3JldApkb21haW4tbmFtZSA9IGRlZmF1bHQKYXV0aC11cmwgPSBodHRwOi8vMTkyLjE2OC4yMDAuOS9pZGVudGl0eQp0ZW5hbnQtaWQgPSBjYzM0YjExZmY5NWQ0MjMwOTA4MWQwYmQ0NmMwZmY4OQpyZWdpb24gPSBSZWdpb25PbmUK
  cloud.conf: W0dsb2JhbF0KdXNlcm5hbWUgPSBhZG1pbgpwYXNzd29yZCA9IFpFMExOdGhIa2dmUzdNMTNLakE2YUJkWjRBMXgzdzNuZE5VS2x1WU8KZG9tYWluLW5hbWUgPSBkZWZhdWx0CmF1dGgtdXJsID0gaHR0cDovLzEwLjE0NS42OS42OjM1MzU3L3YzCnRlbmFudC1pZCA9IDY0NDEzNjdlYzRhZTRjZTM5ZTVmMjY1ZGQwMWRiY2IxCnJlZ2lvbiA9IGJqY2NfcmVnaW9uXzIKCltNZXRhZGF0YV0Kc2VhcmNoLW9yZGVyID0gY29uZmlnRHJpdmUsbWV0YWRhdGFTZXJ2aWNlCg==
  1. 修改k8s master参数
  • 修改kube-controller-manager参数,并重启
  # 启动参数添加两行
  --cloud-provider=openstack \
  --cloud-config=/etc/kubernetes/cloud.conf

如果是kubeadm 以pod的形式启动的kube-controller-manager,修改/etc/kubernetes/manifests/kube-controller-manager.yaml文件就会自动重建pod
注: 官方已经不推荐修改kube-controller-manager的cloud-provider选项了,以后会废弃。而是直接修改kubelet,由于本人没有测试过,大家可以按照官网跑一把,写的很清楚https://kubernetes.io/docs/tasks/administer-cluster/running-cloud-controller/

  1. 部署csi
/home/cloud-provider-openstack/manifests/cinder-csi-plugin
[root@kubernetes-master cinder-csi-plugin]# kubectl apply -f .

查看csi pod

[root@kubernetes-master cinder-csi-plugin]# kubectl get pod -n kube-system 
NAME                                         READY   STATUS    RESTARTS   AGE
coredns-65dbdb44db-56l6t                     1/1     Running   0          13h
csi-cinder-controllerplugin-0                5/5     Running   0          152m
csi-cinder-nodeplugin-6j2l5                  2/2     Running   0          152m
csi-cinder-nodeplugin-fx9zq                  2/2     Running   0          152m
csi-cinder-nodeplugin-pz2dl                  2/2     Running   0          152m
dashboard-metrics-scraper-545bbb8767-5ch8h   1/1     Running   0          13h
kube-flannel-ds-amd64-5qkhq                  1/1     Running   8          14h
kube-flannel-ds-amd64-5tmqv                  1/1     Running   8          14h
kube-flannel-ds-amd64-z9ntk                  1/1     Running   8          14h
kubernetes-dashboard-65665f84db-dgk6g        1/1     Running   0          13h
metrics-server-869ffc99cd-f2b8v              1/1     Running   0          13h

csi driver

[root@kubernetes-master cinder-csi-plugin]# kubectl get csidrivers.storage.k8s.io
NAME                       ATTACHREQUIRED   PODINFOONMOUNT   MODES                  AGE
cinder.csi.openstack.org   true             true             Persistent,Ephemeral   154m

到此cinder csi已经部署好了,下面测试一下

测试

  • 使用官方的examples
[root@kubernetes-master cinder-csi-plugin]# kubectl -f examples/cinder-csi-plugin/nginx.yaml create
[root@kubernetes-master cinder-csi-plugin]# kubectl get pod
NAME     READY   STATUS    RESTARTS   AGE
nginx   1/1     Running   0          135m
  • 查看pvc是否bound
[root@kubernetes-master cinder-csi-plugin]# kubectl get pvc
NAME                    STATUS   VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS          AGE
csi-pvc-cinderplugin    Bound    pvc-18b3ef69-e2ef-4be8-a579-f2b37e1e53f2   1Gi        RWO            csi-sc-cinderplugin   146m
pvc-snapshot-demo       Bound    pvc-ed0de1a3-9a64-4984-912b-0a81d4a3c8e4   1Gi        RWO            csi-sc-cinderplugin   114m
snapshot-demo-restore   Bound    pvc-f6e8e9b2-da5c-45a1-ba33-a230ee9cc5ba   1Gi        RWO            csi-sc-cinderplugin   110m

  • 计入容器查看挂载情况
[root@kubernetes-master cinder-csi-plugin]# kubectl exec -it nginx2 bash
kubectl exec [POD] [COMMAND] is DEPRECATED and will be removed in a future version. Use kubectl kubectl exec [POD] -- [COMMAND] instead.
root@nginx2:/# mount  | grep sdb
/dev/sdb on /var/lib/www/html type ext4 (rw,relatime,seclabel,data=ordered)
root@nginx2:/# ls /var/lib/www/html/
index.html  lost+found  test
root@nginx2:/# echo "test cinder" > /var/lib/www/html/newfile
root@nginx2:/# ls /var/lib/www/html/
index.html  lost+found  newfile  test
root@nginx2:/# 

快照等一些列测试请参考官方文档,文末附链接

参考

https://github.com/kubernetes/cloud-provider-openstack/blob/master/docs/using-cinder-csi-plugin.md
https://github.com/kubernetes/cloud-provider-openstack/blob/master/docs/using-openstack-cloud-controller-manager.md#steps
https://stackoom.com/question/3zJL9/%E5%A6%82%E4%BD%95%E5%B0%86Openstack-Cinder%E5%8D%B7%E7%94%A8%E4%BD%9CKubernetes%E6%8C%81%E4%B9%85%E5%8D%B7
https://kubernetes.io/docs/reference/command-line-tools-reference/kube-controller-manager/

你可能感兴趣的:(k8s集群使用cinder作为后端存储)