add rbac rules for ceph chart

rootfs commented 7 years ago

@alram @dmick PTAL (and test)

alram commented 7 years ago

Just gone one issue creating the storageclass: Error: release ceph failed: User "system:serviceaccount:kube-system:default" cannot create storageclasses.storage.k8s.io at the cluster scope. (post storageclasses.storage.k8s.io)

The rest seems fine.

rootfs commented 7 years ago

that's odd, storageclass is set at cluster scope and bound to kube-system:default

rootfs commented 7 years ago

@alram fixed, storageclass should be clusterrolebinding, thanks for testing!

dmick commented 7 years ago

Man that's a lot of configuration data

rootfs commented 7 years ago

write a debug note here so I can remind me when I forget.

for errors during helm install, like seen by @alram . errors are on the console.
during run time when pod is already created but wants to access resources like endpoint, need to go to the node and see it from docker logs. This is for OSD Pod. There is no other sign, too bad.
could be other errors during helm del --purge, need to see tiller logs, i.e. User "system:serviceaccount:kube-system:default" cannot update daemonsets.extensions in the namespace "ceph"

ceph / ceph-helm

add rbac rules for ceph chart #18