Message boards :
Number crunching :
Results overdue after 4 days?
Message board moderation
Author | Message |
---|---|
cormacolindo Send message Joined: 22 Jan 00 Posts: 9 Credit: 378,023 RAC: 0 |
Hi guys, I just gut no credits for 18 WUs because of "No Reply", ie "too late". But these units were returned only 4 days after they had been sent. All these units were sent on April 4th 2:42 UTC and reported on April 8th 10:52 UTC. This is one example of such a work unit: 122521997 This has to be an error. At least I hope it to be an error, I'd hate to lose the credits for 18 WUs. Has someone else had the same problem? Is there a way to get the credits? |
Keith T. Send message Joined: 23 Aug 99 Posts: 962 Credit: 537,293 RAC: 9 |
Hi guys, You may still get credit for ones that have not reached Quorum like: http://setiathome.berkeley.edu/workunit.php?wuid=122522026 and http://setiathome.berkeley.edu/workunit.php?wuid=122522012. But I think any that are 4th results will not get credit. Sir Arthur C Clarke 1917-2008 |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
Seti Enhanced has been out since May 6th of 2006. Before that the deadlines were a fixed 14 days, but now with the "enhanced" version, they change the deadlines pursuant to the Angle Range of the WU. Deadlines vary between 4 and 55 days. See this post in the "enhanced FAQ" sticky thread. |
bounty.hunter Send message Joined: 22 Mar 04 Posts: 442 Credit: 459,063 RAC: 0 |
Hi guys, In all probability these are "ghost units", i.e., results that appear on seti@home servers but are actually missing from your pc..... These were never reported back, the time you have mentioned as being reported back is actually the deadline for the workunit. Unless of course, you had downloaded and crunched these units but then detached and reattached to seti without reporting the finished workunits first. |
cormacolindo Send message Joined: 22 Jan 00 Posts: 9 Credit: 378,023 RAC: 0 |
bounty.hunter, you are right of course, the WUs have not been reported. However I did not detach/reattach the project either. I have no access to my machines right now, so I can't tell if these are "ghosts". astro, In my opinion, 4 days is an awfully short time to return a result, I use a 3 day cache. |
Idefix Send message Joined: 7 Sep 99 Posts: 154 Credit: 482,193 RAC: 0 |
Hi cormacolindo, In my opinion, 4 days is an awfully short time to return a result, I use a 3 day cache. Your Boinc client takes care that these workunits are reported in time ("Using earliest-deadline-first scheduling because computer is overcommitted" etc.). But your computer has already crunched much younger workunits with much longer deadlines. Indeed, this is a big evidence that these missing workunits are ghosts, . Regards, Carsten |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
If a work unit is returned before the deadline, it will get credit -- even if it is the 4th result, and even if the other work units are returned well before the last one. If the first three do not form a quorum, BOINC will assign new results with a later deadline -- if you get your work in before the new deadline, your late work unit will still get credit. |
KWSN THE Holy Hand Grenade! Send message Joined: 20 Dec 05 Posts: 3187 Credit: 57,163,290 RAC: 0 |
Hi cormacolindo, Also, be aware that work units (WU)'s with only a 4 day deadline will move ahead of units with later deadlines! Look in your "messages" tab and you'll see, (in addition to the message mentioned!) right after the WU downloads, that "the computer is overcommitted"; and that "work fetch is suspended". I agree that 4 days is too short a deadline for these very large angular range (AR) WU's... I think 6-7 days is more like it. (-or whatever doesn't trigger the "overcommited" status!) I keep a 4.5 day cache (for a quite fast computer, thank you!) so I can keep crunching when Berkeley has problems, and these mess up my chances for getting Canonicle results. (the result that is kept by SETI after verification.) I've been known to increase the cache to 9 days when I'm away from home for a few days! If you're gonna be away from home, check for abnormally short "to completion" times in the "tasks" tab - and upload those just b4 you leave! (I'd set the cache to 9 days the day before I leave, and do this final upload [with "no new work" turned on] just before departure... make sure after the WU's are all uploaded that you hit the "update" button to report them!) Of course, ya gotta remember to re-set it back to your usual cache after ya get back... . Hello, from Albany, CA!... |
Challass Send message Joined: 17 May 99 Posts: 7 Credit: 97,559 RAC: 0 |
hello i posted this question on another thread but i think it belongs here. so here we go i am using chickens 2.2b with graphics and every thing was running fine i thought until last night when my results tarted being marked overdo when they were not the last one was sent last 4/12/2007 and was not do until 4/26/2007 what is going on hear? help please thanks |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
I'm not sure I understand your problem here. The only "problem" results I see in your host listing are a few user aborted downloads, and a batch where BOINC detached from SAH for some reason. None have gone "red" at this point which is the online indication of an overdue result. As far as the aborted DL's, perhaps you're intrepreting the reported date as the deadline. For this column "In Progress" is green and displays the deadline, "In Progress, but late" is red and displays the deadline, "Completed" is black and displays the date and time you reported the result. The detached client results are a different matter. Is not all that common but BOINC can end up losing it's host ID have new one spawned from the project, but in this case it looks like you did the detach from BOINC Manager. Keep in mind that these results will go "red" eventually, since a part of the detach process is a project reset and therefore the result input files were dumped. Also, any which were completed but not reported yet were dumped as well. HTH, Alinator |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14653 Credit: 200,643,578 RAC: 874 |
There are a few in his/her result list - like 512527448 (from about 15 pages in :-( ) - which match the thread description. Sent 5 Apr 2007 10:09:24 UTC Received 13 Apr 2007 12:32:15 UTC Outcome Success Report deadline 30 Apr 2007 3:50:58 UTC Validate state Result was reported too late to validate Claimed credit 62.1284673103579 Granted credit 0 Definitely something wrong there, but I don't know if it's possibly related to the detach/attach. |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
OOPS, missed that sorry! :-( That is strange! Maybe this is related to the "Zombie Hunt" Matt's been running this week. He made a post over in Tech News there was a problem and it didn't quite work out as he intended. This isn't the same effect he was talking about but I suppose the errant "dumping" of some WU data could result in this. Alinator |
Challass Send message Joined: 17 May 99 Posts: 7 Credit: 97,559 RAC: 0 |
I'm not sure I understand your problem here. The only "problem" results I see in your host listing are a few user aborted downloads, and a batch where BOINC detached from SAH for some reason. None have gone "red" at this point which is the online indication of an overdue result. |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
OK, the download problem was most likely a result of the "Zombie Hunt" trouble Matt was talking about. You should be able to re-attach now without issues. Generally speaking the worst thing you can do if you're running the recommended version is to just start resetting the project and/or detaching hosts when BOINC falls of out of it's normanl pattern, especially right now as the SAH team is refining their backend and preparing to rollout some new stuff. There was a time not that long ago where this was about the only viable alternative, but that day is past for the most part. For example I have a host which pick up two days of stalled DL's during this event. I didn't touch it and the problem cleared on it's own, without a single lost completed result or a "ghost" DL AFAICT. That's not to say new bugs don't crop up every so often, but again immediately going to a reset or detach may eliminate important information which could help get to the root issue which caused the problem in the first place. In that event it may appear you "solved" the issue, but it could come back to bite you again and you'd be no further along in getting to the real solution than you were the first time. It's almost always better to try and work the problem first. If you need help doing that, that's why a lot of us are here and are more than willing to work with you on resolving the immediate problem and determining the root cause so it can be eliminated in a future build of BOINC and/or the science app. Alinator |
makosky Send message Joined: 7 Jul 00 Posts: 56 Credit: 3,908,782 RAC: 0 |
is boinc having problems iam in class 07 07 2000... when i go to user results it has not changed in 4days and goes for everyone else in that group .. i can upload and d/l fine and it slows my progress in other other areas , but not in my class standing |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
I'm assuming you're talking about the numbers on the third party stat sites, and the reason for that is they haven't made a xml export stat run here since 4/09. Don't worry, when they do you will most likely see a big jump in your numbers after that. ;-) Alinator |
makosky Send message Joined: 7 Jul 00 Posts: 56 Credit: 3,908,782 RAC: 0 |
thank you for your info...at times i wish we still had classic LOL |
haddock29 Send message Joined: 18 Sep 99 Posts: 36 Credit: 26,012,417 RAC: 0 |
There are a few in his/her result list - like 512527448 (from about 15 pages in :-( ) - which match the thread description. I got many of these errors today Look at 517071948, for example: sent apr 12 , result received apr 14, deadline apr 25, and "too late to validate"... In fact, almost all the results uploaded today Apr 14 have that error. But: this is only for one linux box (2*Intel 5160), all my others computers got credits today. Then, why ? Its not sure they have an ET clock in Berkeley, because most computers have credits. May be its the optimized seti client. But i have optimized clients on all the computers, and they should more likely give "invalid", not "too late". Also, the computer date is OK (Apr 14, UT+1) |
haddock29 Send message Joined: 18 Sep 99 Posts: 36 Credit: 26,012,417 RAC: 0 |
I got many of these errors today Look at 517071948, for example: sent apr 12 , result received apr 14, deadline apr 25, and "too late to validate"... In fact, almost all the results uploaded today Apr 14 have that error. But: this is only for one linux box (2*Intel 5160), all my others computers got credits today. Then, why ? Its not sure they have an ET clock in Berkeley, because most computers have credits. May be its the optimized seti client. But i have optimized clients on all the computers, and they should more likely give "invalid", not "too late". Also, the computer date is OK (Apr 14, UT+1) [/quote] Sorry, a copy glitch. The example was: http://setiathome.berkeley.edu/result.php?resultid=515699866 |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14653 Credit: 200,643,578 RAC: 874 |
I got many of these errors today Look at 517071948, for example: sent apr 12 , result received apr 14, deadline apr 25, and "too late to validate"... No, you were right the first time - result 517071948 is the example. This machine is one of the ones which has suffered 'client detached' - I think this is far more likely to be the cause of the 'too late to validate' error, rather than optimisation or a faulty RTC. Maybe the error message should be read as (or amended to read) 'Result reported after client detached'. |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.