kubernetes - 当 heketi 端点与 pv 和 PVC 不在同一个命名空间时,glusterfs 如何创建卷
问题描述
我有两个命名空间“runsdata”和“监控”。heketi pod 和 glusterfs 的 daemonSet pod 都在“runsdata”命名空间下。现在我想在 'monitoring' 命名空间下创建 Prometheus 监视器。因为我需要存储来存储我的 Prometheus 数据。所以我创建了PVC(在'monitoring'ns下)和pv,并在PVC yaml中声明了storageclass来创建相应的卷,以便为Prometheus提供存储。但是当我创建与 pv 绑定的 pvc 并应用 prometheus-server.yaml 时。我得到错误:
Warning FailedMount 18m (x3 over 43m) kubelet, 172.16.5.151 Unable to attach or mount volumes: unmounted volumes=[prometheus-data-volume], unattached volumes=[prometheus-rules-volume prometheus-token-vcrr2 prometheus-data-volume prometheus-conf-volume]: timed out waiting for the condition
Warning FailedMount 13m (x5 over 50m) kubelet, 172.16.5.151 Unable to attach or mount volumes: unmounted volumes=[prometheus-data-volume], unattached volumes=[prometheus-token-vcrr2 prometheus-data-volume prometheus-conf-volume prometheus-rules-volume]: timed out waiting for the condition
Warning FailedMount 3m58s (x35 over 59m) kubelet, 172.16.5.151 MountVolume.NewMounter initialization failed for volume "data-prometheus-pv" : endpoints "heketi-storage-endpoints" not found
从上面的日志不难知道,storageClass找不到heketi端点来创建volume。因为 heketi 端点位于“runsdata”下。我怎么解决这个问题?
其他信息: 1. pv 和 pvc
apiVersion: v1
kind: PersistentVolume
metadata:
name: data-prometheus-pv
labels:
pv: data-prometheus-pv
release: stable
spec:
capacity:
storage: 2Gi
accessModes:
- ReadWriteMany
persistentVolumeReclaimPolicy: Recycle
storageClassName: runsdata-static-class
glusterfs:
endpoints: "heketi-storage-endpoints"
path: "runsdata-glusterfs-static-class"
readOnly: true
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: data-prometheus-claim
namespace: monitoring
spec:
accessModes:
- ReadWriteMany
resources:
requests:
storage: 2Gi
storageClassName: runsdata-static-class
selector:
matchLabels:
pv: data-prometheus-pv
release: stable
[root@localhost online-prometheus]# kubectl get pv
NAME CAPACITY ACCESS MODES RECLAIM POLICY STATUS CLAIM STORAGECLASS REASON AGE
data-config-pv 1Gi RWX Retain Bound runsdata/data-config-claim runsdata-static-class 5d22h
data-mongo-pv 1Gi RWX Retain Bound runsdata/data-mongo-claim runsdata-static-class 4d4h
data-prometheus-pv 2Gi RWX Recycle Bound monitoring/data-prometheus-claim runsdata-static-class 151m
data-static-pv 1Gi RWX Retain Bound runsdata/data-static-claim runsdata-static-class 7d15h
pvc-02f5ce74-db7c-40ba-b0e1-ac3bf3ba1b37 3Gi RWX Delete Bound runsdata/data-test-claim runsdata-static-class 3d5h
pvc-085ec0f1-6429-4612-9f71-309b94a94463 1Gi RWX Delete Bound runsdata/data-file-claim runsdata-static-class 3d17h
[root@localhost online-prometheus]# kubectl get pvc -n monitoring
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
data-prometheus-claim Bound data-prometheus-pv 2Gi RWX runsdata-static-class 151m
[root@localhost online-prometheus]#
- heketi 和 glusterfs
[root@localhost online-prometheus]# kubectl get pods -n runsdata|egrep "heketi|gluster"
glusterfs-5btbl 1/1 Running 1 11d
glusterfs-7gmbh 1/1 Running 3 11d
glusterfs-rmx7k 1/1 Running 7 11d
heketi-78ccdb6fd-97tkv 1/1 Running 2 10d
[root@localhost online-prometheus]#
- 存储类定义
---
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
name: runsdata-static-class
provisioner: kubernetes.io/glusterfs
allowVolumeExpansion: true
reclaimPolicy: Delete
parameters:
resturl: "http://10.10.11.181:8080"
volumetype: "replicate:3"
restauthenabled: "true"
restuser: "admin"
restuserkey: "runsdata-gf-admin"
#secretNamespace: "runsdata"
#secretName: "heketi-secret"
解决方案
解决方案是在当前命名空间下创建端点和服务。然后我们可以在 pv yaml 中使用该服务,如下所示:
[root@localhost gluster]# cat glusterfs-endpoints.yaml
---
kind: Endpoints
apiVersion: v1
metadata:
name: glusterfs-cluster
namespace: monitoring
subsets:
- addresses:
- ip: 172.16.5.150
- ip: 172.16.5.151
- ip: 172.16.5.152
ports:
- port: 1
protocol: TCP
[root@localhost gluster]# cat glusterfs-service.yaml
apiVersion: v1
kind: Service
metadata:
name: glusterfs-cluster
namespace: monitoring
spec:
ports:
- port: 1
[root@localhost gluster]#
推荐阅读
- python - 如何使用多个正则表达式函数在python中提取数据
- javascript - 在for循环丢失索引中使用对话框时出现VUETIFY问题
- slack - 我的 slackapp 不适用于其他工作区
- cassandra - 设置和启动 cassandra 进行生产的选项
- amazon-web-services - AWS ECS Fargate 内存使用情况
- mysql - 使用 NOT IN 对具有奇怪的空值的列表的 SQL 查询
- javascript - Discord.js 我无法创建具有特定权限的角色,该角色是静音角色
- java - 在 Spring 应用程序中运行内存中(嵌入式)MongoDB 的集成测试
- amazon-web-services - AWS Cloudfront - 具有 CloudFront 源访问和引用存储桶策略的 S3
- python - Selenium 的 Youtube 网页抓取