Ceph on Kubernetes & RC 控制每一个节点上OSD的数量

[jiamliang]

1. 在k8s中运行ceph
2. k8s中如何控制每一个node上，有与硬盘数量对应个数的osd（系统盘除外）

[jiamliang]

按照这篇文档操作：

liangjiameng@liangjiameng-Ubuntu:kubernetes$ kubectl get all --namespace=ceph
NAME                            CLUSTER-IP   EXTERNAL-IP         PORT(S)    AGE
ceph-mon                        None         <none>              6789/TCP   40s
NAME                            READY        STATUS              RESTARTS   AGE
ceph-mds-1090156222-g6s0q       0/1          Pending             0          40s
ceph-mon-2254771767-kj9e1       0/1          ContainerCreating   0          40s
ceph-mon-2254771767-qs9ys       0/1          ImagePullBackOff    0          40s
ceph-mon-2254771767-u9o5n       0/1          ContainerCreating   0          40s
ceph-mon-check-76715991-yadqg   0/1          ImagePullBackOff    0          40s
liangjiameng@liangjiameng-Ubuntu:kubernetes$ kubectl get all --namespace=ceph
NAME                            CLUSTER-IP   EXTERNAL-IP        PORT(S)    AGE
ceph-mon                        None         <none>             6789/TCP   51s
NAME                            READY        STATUS             RESTARTS   AGE
ceph-mds-1090156222-g6s0q       0/1          Pending            0          51s
ceph-mon-2254771767-kj9e1       0/1          ErrImagePull       0          51s
ceph-mon-2254771767-qs9ys       0/1          ImagePullBackOff   0          51s
ceph-mon-2254771767-u9o5n       0/1          ErrImagePull       0          51s
ceph-mon-check-76715991-yadqg   1/1          Running            0          51s

应该是docker image pull下来失败了（新建的集群没有翻墙功能）。

TODO:
准备按照@pineking 的方法，先docker pull ceph/daemon ，再push到内网私有docker hub。
在yaml中指定从私有的docker hub去pull。