I tested this with postgresql and noticed that the default charmhelpers mount creation algorithm has a bug with permissions and ownership. I'll file a bug on that separately, but that led me to think that we should be creating the directory in the local provider so it is ready to go for the caller.
a simple 'os.makedirs()' on the mountpoint if it does not exist would be helpful and expected IMO.
Other than this, things look great. I'll mark needs fixing until then.
I tested this with postgresql and noticed that the default charmhelpers mount creation algorithm has a bug with permissions and ownership. I'll file a bug on that separately, but that led me to think that we should be creating the directory in the local provider so it is ready to go for the caller.
a simple 'os.makedirs()' on the mountpoint if it does not exist would be helpful and expected IMO.
Other than this, things look great. I'll mark needs fixing until then.