Folding Forum

Yup, you read right, that WU infuriatingly EUE'd at 95%, to make matters worst, it took me 6 days to notice what was going on :

P2665 R0 C110 G27, DL'ed at 6h11 on the 9th, EUE'd at 95% at 6h00 on the 10th
P2665 R0 C110 G27, DL'ed at 6h09 on the 10th, EUE'd at 95% at 6h00 on the 11th
P2665 R0 C110 G27, DL'ed at 6h12 on the 11th, EUE'd at 95% at 5h46 on the 12th, triggered a core DL
P2665 R0 C110 G27, DL'ed at 5h49 on the 12th, EUE'd at 95% at 5h34 on the 13th
P2665 R0 C110 G27, DL'ed at 5h34 on the 13th, EUE'd at 95% at 5h28 on the 14th
P2665 R0 C110 G27, DL'ed at 5h30 on the 14th, EUE'd at 95% at 5h24 on the 15th, triggered a core DL
P2665 R0 C110 G27, DL'ed at 5h27 on the 15th, Caught and manually purged
P2665 R0 C110 G27, DL'ed at 5h33 on the 15th, Caught and manually purged

Then the assignment server finally gave up trying to send me that bletcherous WU.

This machine is 100% stable, the log shows that 14 others 2653 & 2665 were completed before the R0 C110 G27 "infection".

For those interested, you can see the unadulterated 700KB logfile Here

I am not amused, I lost well over ten thousand points in the last few days due to that crap WU and the assignment server's bad habit of re-issuing the same bad WU to the same user over and over again.

While I could have caught it earlier, the EUE's happened so close to the WU end that I didn't notice anything weird, the % completed I observed when checking on the client were consistent with the client having started a new WU after the previous one finished.

I don't mind being a bit more attentive to avoid that spurious behavior but I'm not interested in babysitting the client, especially when the assignment server just insist on ruining almost a week worth of processing and massive point lossage.

Upgrade your client to 6.23, it solves this kind of issues (it can detect EUE and send the partial result back) : viewtopic.php?f=46&t=6642

Thanks, will drop in the 6.23 executable as soon as I get back home, any hope to recover any points ? I made a backup of the work folder & queue.dat file before repeatedly purging that crap WU.

You can try qfix : viewtopic.php?f=19&t=6042

Just in case you skimped over the first post, I am talking about the same WU that was processed six times up to 95% completion before EUE'ing, obviously, qfix will only yield partial points for a single queue slot, as resubmitting what's left in the other slots will just be discarded as duplicates by the collection server without earning me any points, right ?

You'll get credit only once ... but the server will know that the WU is bad and won't send it to you again.

Someone else submitted multiple results for this WU ... they were all for 0 points.

Considering the deadline expired and the WU was reissued to someone else, I lost a week crunching that WU for 0 points, right ?

Not happy...

For the record, using qfix does not tell the server a WU is bad. It only prevents you from getting reassigned it right away. The problem reported by the OP and the incomplete nature of the qfix patch are why we recommend using 6.23 (especially for all SMP clients).

95% before it EUE'd six times in a row is just cruel, thank you very much !

I'm really looking forward to a WU that craps out at 99% and tie up my client for a frickin' week and a half, that'd be hardcore !

SidVicious wrote:Considering the deadline expired and the WU was reissued to someone else, I lost a week crunching that WU for 0 points, right ?

Not happy...

It should be noted that you did not produce any useful science during that week either. LIfe isn't always fair, but in this case, it's at least rational.

Nevertheless, I wouldn't be happy, but I check on my progress more frequently than that and I'm sure I don't need to remind you that all beta software needs to be watched. Of course I have already upgraded to 6.23, so it wouldn't have happened to me.