The last-commit-doc logic should have loop detection and a configurable timeout so that worst-case scenarios cannot tie up system resources
I'm not sure this ticket should be re-opened. Unless we're sure there's a bug in this change, the issue remaining in production means the problem wasn't in the LCD logic after all. But I guess we can leave it open until we isolate the exact issue with verbose TMW logging.
How wrong I was. :-) Fix for the timeout force-pushed to allura:cj/6695