Thanks for your reply, Alex. The answer to your question is pretty bizarre: My watches had been running (though at strangely irregular intervals as I mentioned earlier), and in fact the reason that I tried to delete one of them was because it was alerting me in situations when I didn't want it to. So, clearly Watcher was running, and thus it never occurred to me to check whether it was running. (Being kind of a novice, I assumed that the "not started" part of the error was referring to the watch, not Watcher -- which still didn't make any sense, but anyway...) But after receiving your reply, I checked Watcher's status, and sure enough, it said it's not running:
{ "watcher_state" : "stopped", "watch_count" : 18, "execution_thread_pool" : { "queue_size" : 0, "max_size" : 20 }, "manually_stopped" : false }
So I tried starting it, and that gave me:
failed to start watcher. please wait for the cluster to become ready or try to start Watcher manually
So then I restarted ES (which has been working fine), and I got that same "failed to start" message again. And yet my watches are still running (although irregularly).
Also -- and I don't know if this is related -- the license information during startup said that our license expires on August 10, so somehow it reverted from our license which is good until next April to a 30-day trial license. I was able to reinstall our license successfully. But things aren't working any better.
So I'm lost. I keep thinking that maybe some stuff got corrupted when the server ran out of space recently.
Also, I now see that we're sometimes getting errors saying that a watch failed to run due to thread pool capacity. I can't seem to find any useful information about that.
Thanks for any further help you can provide.