Was previously kubernetes_auth format.
Added defaults file which uses salt to fill in an env var
with the master's IP.
More thought needs to be given soon to how to make this
connection use a cert for the master, and how to support
multiple masters, and whether to use the DNS record
instead of an IP address. But this PR unblocks some other
more urgent things, so doing it this way.
These secrets will be used in subsequent PRs by:
scheduler, controller-manager, monitoring services,
logging services, and skydns.
Each of these services will then be able to stop using kubernetes-ro
or host networking.
The better solution is some fence with Salt, but the actual logs
provided in the bug don't support any race condition here, plus the
ordering in the Salt configuration seems correct.
We haven't seen this again in a while, but given the results of the
situation (a borked cluster), I'm proposing a relatively simple
workaround.
Fixes#4357 (dubiously)