First New Results INVALID???
18 Jan 2007 23:07:11 UTC
Topic 192332
(moderation:
I've got several new results that are pending, awaiting the return of the second one to form the quorum. This one - 26343979 has now been tested and is listed as invalid. Looks like we have a problem with the validator.
Normally you would expect a mismatch between results to be determined by sending out a third one to decide the issue. On this occasion, both have been immediately marked as invalid and two new results are listed to be sent out. There is nothing a user can do except wait for the Devs to find what is causing this. I'm confident it will get resolved fairly promptly once they work out the cause. No need to panic - yet :).
UPDATE: I am now seeing new S5RI results which are validating and being awarded credit. Looks like Bruce has removed the rogue validator and replaced it with the correct one.
Cheers,
Gary.
First New Results INVALID???
)
That looks strange, just two examples :
Spindown-range [-7.672054e-09, 7.672060000000007e-10] (Linux)
Spindown-range [-7.672054e-009, 7.672060000000007e-010] (Windows)
The validator needs to know that those two values are identical, plain binary comparison will not do.
p.s.: from http://einsteinathome.org/node/192312 by BM, posted just now
That's not good. I think I'll
)
That's not good. I think I'll suspend until that gets fixed because just checked the work I have and my qorum mates are on linux and i'm on windows. I have one that has reported already but does not have a mate to validate against yet. Will wait to see what happens to that one before I decide if I will do more.
Steve
98SE XP2500+ @ 2.1 GHz Boinc v5.8.8
RE: p.s.: from
)
Ahh... Thanks for that. Good to see Bernd is on to it already.
Now, how do we stop a million people bitching about their results being marked INVALID!!! :).
Cheers,
Gary.
RE: That's not good. I
)
I wouldn't worry about it too much. From the tone of Bernd's message, Bruce will wave his magic wand and it will all get sorted anyway.
Cheers,
Gary.
RE: Now, how do we stop a
)
Well it's an interesting ratio, the number of registered/active users/machines ( ~ hundreds of thousands ) vs. the number of workers maintaining the project ( ~ half a dozen ). It's biggie and certainly their dedication and effort is reflected in the small amount of down time and difficulty we've had over the whole course of the project.
Still one can't account for all tastes, and personally I think the recent difficulties have exposed more about some participants than about any supposed laxity of the E@H staff. :-)
As I've indicated before, when the gravity waves are confirmed, the world will be crawling all over this site. So remember dear colleagues, it's ever so much harder to retract than to issue complaint.
One should never, of course, be reticent in providing polite & healthy feedback.
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
I've added an update to the
)
I've added an update to the first post in this thread as I'm now seeing new S5RI results which are corectly validating. Hopefully this signals the end of the problem.
Cheers,
Gary.
It also looks like the ones
)
It also looks like the ones that were invalid were fixed. I had a couple before I went to bed last night, now those are corrected and have their points.
RE: It also looks like the
)
I got a special WU where I got no credits at all... :-(
My assumption:
The first two results got status 'invalid' because of the validation error.
The WU was sent out 2 more times.
The other two hosts returned a result which got validated.
Bruce (or Bernd) corrected all WUs with validation error but at this time the quorum was already built and the first two WUs were to late...
Bernd, is there any chance to get the credits nevertheless?
Udo
RE: Bernd, is there any
)
We're looking into this. Might be we don't find a suitable solution (that keeps the DB consistent) today anymore, but thanks for pointing us to the problem.
BM
BM