While preparing to update a test cluster, I accidentally attempted to refresh some monitor pages after I had shut down the underlying HDFS and Zookeeper services. The page simply hung. I've since tried loading several different pages (overview, logs, master view) and all simple hang on the monitor's end. If I restart the underlying cluster, the pages resolve like normal.
I haven't had a chance to see if this happens on later versions, nor figure out which missing component matters.
Expected behavior would be for the monitor to still load, albeit with a bunch of alarm bells about everything being off.