If it's a Stale, then KanoDB will mark it as an Orphan almost immediately.

There's a few bitcoind's on the main server, and KanoDB talks to a different one to the one the pool uses to generate work.
Thus if a Stale occurs and ckpool tells it to switch to the Stale block, the others wont switch, since they'd have to be told to, and ckpool can only talk to one of them, and KanoDB will see that the block isn't in the blockchain according to it's bitcoind and thus flag it as Orphan.

In the case of a true Orphan, we'd not know about it until the block is actually orphaned by the block after it, since before that it isn't an orphan.
If the re-org happens, KanoDB will mark it as an Orphan due to the bitcoin re-org saying that the block we found is no longer in the blockchain.

I can then manually change the state if necessary, but KanoDB automatically sets them to Orphan if they aren't in it's blockchain.