Questions and Answers :
Windows :
inconclusive work
Message board moderation
Author | Message |
---|---|
Send message Joined: 4 May 11 Posts: 4 Credit: 206,427 RAC: 0 |
please take a look at this wu: http://moowrap.net/workunit.php?wuid=26035 it's completed successfully by another system. Is there credit granted sometime or are wus like this staying "inconclusive" forever? thanks for an answer |
Send message Joined: 2 May 11 Posts: 2 Credit: 66,286,394 RAC: 0 |
I was just going to start a new thread on this problem, glad to see I'm not alone. :) On my slower GPUs (9800GT) the version 1.00 Wus returned the last couple of days are showing “validation inconclusiveâ€, 1.01 WUs from the same machines appear to be validating alright. I aborted the remaining 1.00 work queued on those machines. Are these results actually bad or was something changed on the Validator to cause the older 1.00 WUs to be marked as inconclusive? http://moowrap.net/workunit.php?wuid=30516 wuid=29238 wuid=28523 wuid=28027 wuid=30316 wuid=27718 wuid=27223 wuid=26477 wuid=25539 wuid=25354 wuid=25208 wuid=24620 wuid=24531 wuid=24530 wuid=24841 wuid=24801 wuid=24793 wuid=23594 wuid=23574 wuid=22938 wuid=22214 wuid=22007 wuid=21905 Sorry for the lack of links, the board SPAM filter would not let me post that many URLs :( |
Send message Joined: 4 May 11 Posts: 4 Credit: 206,427 RAC: 0 |
I was just going to start a new thread on this problem, glad to see I'm not alone. :) you're definitly not alone ;-) you're right it seems: just checked and the wus i'm (we're) talking about are 1.00 version. thanks for info, didn't see that |
Send message Joined: 2 May 11 Posts: 65 Credit: 242,754,987 RAC: 0 |
yes, Teemu made a change in the validation (regarding this problem) for version 1.01. If you guys see that same problem with 1.01, that's definitly a problem. |
Send message Joined: 8 May 11 Posts: 11 Credit: 1,075,941 RAC: 0 |
Another one: http://moowrap.net/workunit.php?wuid=54907 http://moowrap.net/result.php?resultid=62471 |
Send message Joined: 2 May 11 Posts: 65 Credit: 242,754,987 RAC: 0 |
good catch. Sadly, Teemu will have to answer to this one. I also noticed, i've few inconclusive too: http://moowrap.net/results.php?userid=616&offset=0&show_names=0&state=2&appid= Also, what is inconclusive work really means? Another bug with the validator? |
Send message Joined: 20 Apr 11 Posts: 388 Credit: 822,356,221 RAC: 0 |
Hi, That validator change (basically it's more strict now) affects all workunits. But older application version is more susceptible for getting caught from having problems. This means that the result was not fully done by the client and validator detected that. That's why there's a need to redo workunit with a new result. Once that other one is done and validated, the inconclusive result is supposed to get partial credit. But it seems validator isn't granting partial credit correctly at the moment. I need to fix that but rest assured that partial credit for those results will be granted. :) -w |
Send message Joined: 4 May 11 Posts: 4 Credit: 206,427 RAC: 0 |
thanks for your efforts |
Send message Joined: 8 May 11 Posts: 11 Credit: 1,075,941 RAC: 0 |
I need to fix that but rest assured that partial credit for those results will be granted. :) Thank you. |
Send message Joined: 20 Apr 11 Posts: 388 Credit: 822,356,221 RAC: 0 |
Hi, Okay, partial crediting should be operational. Note that it can take up to a day from the time inconclusive result is detected to actually getting the credit. And longer if getting back a valid result is delayed. -w |
Send message Joined: 2 May 11 Posts: 27 Credit: 1,151,788 RAC: 0 |
i got one WU which was marked as invalid, but stderr output does not show any strange things: http://moowrap.net/result.php?resultid=50160 what went wrong? |
Send message Joined: 20 Apr 11 Posts: 388 Credit: 822,356,221 RAC: 0 |
Hi, That result lost one packet and therefore didn't do the full work and got granted only partial credit. Dnet client sometimes looses work like this due to some bug in it. :( -w |
Send message Joined: 2 May 11 Posts: 27 Credit: 1,151,788 RAC: 0 |
Hi, yup, i saw it got credit by now. btw.: they say they have fixed some things in v 519 meanwhile.. |
Send message Joined: 2 May 11 Posts: 65 Credit: 242,754,987 RAC: 0 |
518 fixes something regarding the OGR-27, which is not running on GPU, so it didnt affect us. More infos at: http://blogs.distributed.net/2011/04/11/15/12/mikereed/ But we're already running 518. 519 is in beta stage at the moment for GPU. Like you can see at: http://www.distributed.net/Download_clients |
Send message Joined: 6 May 11 Posts: 15 Credit: 692,725,672 RAC: 645 |
Is this a good example? http://moowrap.net/workunit.php?wuid=88050 My machine produced the canonical fortunatly! |
Send message Joined: 4 May 11 Posts: 7 Credit: 1,744,262,645 RAC: 0 |
|
Send message Joined: 20 Apr 11 Posts: 388 Credit: 822,356,221 RAC: 0 |
All 3 are inconclusive. Aha! Those three indicate a bug in our validator. I'll take a look and hopefully will be able to fix this problem. Thanks for bringing these work units to my attention! -w |
Send message Joined: 16 May 11 Posts: 1 Credit: 24,472 RAC: 0 |
I also get some wus with the status "Completed, validation inconclusive" and later "Completed, marked as invalid". I compared the stderr output of all wus I worked on and I figured out that the ones where I or the other user restored the workstatus from a checkpoint after a restart all became invalid", whereas the wus which run all at once were validated sucessfully. I just finished a few wus, therefore it isn't representative but maybe this is one of the problems why results get invalid. |
Send message Joined: 20 Apr 11 Posts: 388 Credit: 822,356,221 RAC: 0 |
I also get some wus with the status "Completed, validation inconclusive" and later "Completed, marked as invalid". Some inconclusive work and eventual invalid (with partial credit granted) are normal. This happens if the work unit was not fully crunched but got some packets were lost due to some errors. Looking at your three invalid results, all misplaced one packet so that's this normal situation. However, why they loose one packet is something that can be looked into and hopefully fixed. Checkpointing should preserve any packets while wrapper/client is been restarted. Any whole system hang (BSOD on windows) that results in loosing checkpoint files can have this effect. -w |