Any time I run a cloud-wide cumin action without an explicit timeout, it hangs forever due to being unable to reach things in the 'contintcloud' project. VMs there are transient testing instances which aren't fully puppetized so will never have the keys necessary to support cumin access (and, anyway, we never care about them since they're short-lived.)
Similarly, the admin-monitoring project is only used for transient testing VMs.
Can we just have cumin ignore those projects now and forever?