The recent outages appear to have been caused by a bunch of automated hits on the server that overwhelmed it. The source seems to be a somewhat random list of IPs, so blocking those IP's would likely not do much good in preventing a repeat. The fix here is a modification on the software to disregard any requests for thread pages if a huge number of such requests come in a very short time (more than 10 in one second, for example) but such a modification will need to be added to my to-do list, so there may be more such outages in the near future. Until then all I can do is deal with them when they happen. This looks like a Denial of Service (DOS) attack but simple incompetence by some web crawling programmer can't be ruled out. If the cause is incompetence and not a deliberate, malicious attack, then I'd expect this problem to go away.
Thanks for your patience.