I managed to provoke another hang.

[tharrison1]

I managed to provoke another hang. I moved roughly 150 files from the download directory to elsewhere in my library using Windows Explorer. When I was done I went to Shareaza; it was responsive and was hashing "88 files" (apparently it was hashing them about half as fast as I was categorizing and moving them). Without doing anything else to it except check on it occasionally, I got it to hang. I checked on it at one point and it showed 29 files remaining and a busy cursor, of which the latter was a new development. I found it completely unresponsive and again using up a lot of CPU. The latter is normal during hashing anyway, but it wasn't hashing -- the number of remaining files held steady at 29. All progress had stopped.

And the files I'd moved were a maximum of a few megs each. None were larger than about 10MB. It's not that I moved many small files and one 6 GB archive of ISO images or something and it had hit that one very large file and was grinding away at it. And the user interface shouldn't have wedged even if it had hit a 6 GB file while hashing.

I didn't need to manually restart it -- as the previous time, it recovered on its own after about ten minutes, resumed hashing the files, and was shortly done.

So, every so often, Shareaza 2.5.3.0 just stops dead in its tracks and "thinks", to the exclusion of all else (including interface response), for several full minutes before suddenly getting back on track. A human being exhibiting such behavior might be called absent-minded, or (depending on how severe it was) even diagnosed with petit mal epilepsy. I don't think software is supposed to exhibit such complex failure modes, though. Especially not epilepsy, which requires a very different underlying hardware architecture.

Further, it looks like 2.5.3.0 has these spells roughly every six hours in the absence of particularly identifiable provocations (user inputs that could be expected to cause it stress, such as making it hash a very large file or starting a very large and fast torrent download). Either it does it spontaneously, for all intents and purposes, or the triggering stressor is not a user input but a network input. (I've forced my copy not to ever go into hub mode so it isn't that.)

If it's the latter, we're looking at security implications; Shareaza can be sent a "packet of death" that will at least temporarily render it unusable. If it makes it useless for five minutes, sending a "packet of death" every five minutes with a long timeout to a known Shareaza host will at the very least lock that machine's user out of Shareaza and in some cases (laptops) may cause worse problems (rapid battery depletion, overheating) due to the way the hypothetical "packet of death" makes Shareaza busy-loop and overtax the CPU.

I think this warrants looking into by people that know how Shareaza works under the hood. Particularly to see if there's a signal that can be sent to a running copy of Shareaza 2.5.3.0 over the Internet that will trigger it, if you know the port Shareaza's using and the host's IP address.

[kevogod]

Please create your own threads for separate issues. (This thread has been split from http://sourceforge.net/apps/phpbb/share ... ?f=3&t=850.)