THANKS NELS, I have too much time on my hands

Most people have a long holiday weekend and deserve it.

Hi Meshmesh, Mike, et al.

I'm curious about meshmesh's team result problems and we haven't heard back since the initial post?

Mike, I think that your problem will be the power outages. If BOINC is writing a checkpoint when you have a power outage it will return an invalid result. Four invalids from all of your machines during a power outage would seem typical I should think.

That doesn't answer the original question from meshmesh and team unless they all live in the same power district though.

Cheers. ozylynx

Hi meshmesh. You forgot to close the quote, that's why there's a gray bar down the side of the page. Can you edit it please?

Sometimes you get logged out and can't edit. Log back in, and the pencil icon will be back. When you edit a post, it's the "Edit a post" button that submits your edit. Weird, but there you go.

hi mike, just a few thoughts to add to yours



I know nothing about Leiden but I am pretty sure QMC does not use a quorum to validate results. In other words, that computer might have been returning an occassional "bad" result but QMC never told you and just gave you the credits you claimed. By "bad" I mean a result that processed without BOINC detecting any errors on your end but which had errors nevertheless. BOINC cannot detect all errors and that's why some projects use a quorum. Some projects, and I think QMC is one, are able to use methods other than a quorum and some are simply able to use results that have minor errors. Perhaps WCG has more stringent error checking than you are used to so now you're seeing more invalid results than you are used to.

There is no doubt you're using good quality gear but even good quality stuff can deteriorate to the point where fails occassionally under extreme conditions. Crunching FAAH is definitely an acid test.



Defective FAAH workunits have been very few and very far between. By defective I mean a work unit that fails on every machine that crunches it. The work units you are asking about are not defective, there was an error during crunching, an error BOINC did not detect, an error that showed up when compared with the other results in the quorum.



That's the spirit. In my experience some problems just go away on their own and you never find the cause. Other problems take a long time and special tools to track down. Some can be fixed by a lengthy and sometimes expensive process of elimination where you swap out various components until you get lucky and swap out the one causing the problem and then you get to say, "Aha! It was the @$#! torque rotor. I had a hunch it was the torque rotor." The worst is where you actually have 2 bad components but you're convinced you have only 1 bad.



Yah, it is a pretty low invalid rate, if I'm reading your numbers correctly. And it might get even better if you wait a bit.



That is definitely a possible cause. The best quality system in the world will not crunch properly if it isn't powered properly. It's not just power outages you need to worry about, you also need to handle fluctuations in the voltage. All of your machines need to be protected by a good quality spike protector and noise filter. If they are not then the components inside the box will deteriorate and you'll get errors. If you can't afford a UPS then at least get spike protection and noise filters and make sure all of your mains receptacles are grounded properly else the spike protection and noise filters will not work optimally.

---