Message boards :
Number crunching :
False entry in the database?
Message board moderation
Author | Message |
---|---|
zhang Send message Joined: 8 Dec 99 Posts: 2 Credit: 1,006,571 RAC: 0 |
The error message shown in result ID 100140931 has computer ID corresponding to mine but the error log clearly has nothing to do with mine. Any ideas? |
Ingleside Send message Joined: 4 Feb 03 Posts: 1546 Credit: 15,832,022 RAC: 13 |
The error message shown in result ID 100140931 has computer ID corresponding to mine but the error log clearly has nothing to do with mine. Any ideas? Depending on which error happens, some debug-info can be included in stderror-out, and your particular error just shows the full path and name of one of the many source-files on the machine this particular build of seti@home was compiled on, and on which line of this source-file the error happened. Rom Walton is one of the main BOINC-programmers and is normally responsible for new client-releases, so his name showing up on the machine responsible for compiling SETI@Home shouldn't be unexpected. :) |
Don Erway Send message Joined: 18 May 99 Posts: 305 Credit: 471,946 RAC: 0 |
I just got something that *may* be a false entry in the database, or else an error in processing. Very strange. Check out the WU: http://setiathome.berkeley.edu/workunit.php?wuid=22647486 It was all settled, on Aug 4, and the deadline was Aug 17, and it was obviously all done, had 3 good results, and had nothing to do with me. Then, on the 18th, it was sent out again, to one of my PCs, and reprocessed.. And instant credit given. This strikes me as a possible clue, to files getting lost. Any ideas? Don |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
Check out the WU: It's possible and frequent that results are returned "successful" but don't match other returns. It's possible one or two of the three didn't match closely enough and the validator had that one sent out to you and your return helped it decide. |
Don Erway Send message Joined: 18 May 99 Posts: 305 Credit: 471,946 RAC: 0 |
Check out the WU: Hi Tony. But they all 3 got credit, in the end. So doesn't that imply they were all close enough to validate? What would be the need for a 4th result, in that case? And, if these are resent out, how come the WUs with 2 good results, and all the rest, "client download error", are not also resent? Thanks! Don |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
Result ID click for details Computer Sent Time reported or deadline explain Server state explain Outcome explain Client state explain CPU time (sec) claimed credit granted credit 95658971 1199475 3 Aug 2005 12:51:19 UTC 4 Aug 2005 1:01:21 UTC Over Success Done 16,094.88 33.74 20.57 95658972 1196607 3 Aug 2005 12:51:37 UTC 17 Aug 2005 12:51:37 UTC Over No reply New 0.00 --- --- 95658973 602562 3 Aug 2005 12:51:30 UTC 4 Aug 2005 23:58:51 UTC Over Success Done 7,926.92 23.17 20.57 95658974 139682 3 Aug 2005 12:51:16 UTC 4 Aug 2005 13:43:20 UTC Over Success Done 15,954.83 17.98 20.57 102734330 1161224 18 Aug 2005 14:05:03 UTC 18 Aug 2005 23:55:42 UTC Over Success Done 5,295.64 16.85 20.57 OK, here's my guess. The validators are overloaded and didn't get to this one prior to the deadline. The existance of the late WU caused yours to be sent out prior to validation and so you also got credit? I.E if the validators got to it prior to it being resent out then it might not have been resent. |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
I.E One result is resent for every faulty WU, unless stopped before hand by the generation of a canonical result. |
Don Erway Send message Joined: 18 May 99 Posts: 305 Credit: 471,946 RAC: 0 |
I.E One result is resent for every faulty WU, unless stopped before hand by the generation of a canonical result. So, it was considered faulty, when it had not been validated, by the deadline..?? Weird, that it would not just try validating, since it had 3 good results. Something I don't understand here, about the way the servers work. But, if this is the case, things are about to explode, as the validator is frequently going to miss deadlines now, and millions more results will be sent out, and millions more files generated, slowing the validators even further! Don |
John McLeod VII Send message Joined: 15 Jul 99 Posts: 24806 Credit: 790,712 RAC: 0 |
It is also possible that the last result had science values between the two outliers, and thus everything was close enough to the last returned result, but not to each other. BOINC WIKI |
Ingleside Send message Joined: 4 Feb 03 Posts: 1546 Credit: 15,832,022 RAC: 13 |
http://setiathome.berkeley.edu/workunit.php?wuid=22647486 There's two possibilities, either the Validator was so backlogged that it wasn't validated before your copy was sent out. Granted the validator-queue is huge, but it's still not 13-day backlog so this shouldn't be the reason. The other possibility on the other hand, is that the 3 results failed validation with "No consensus yet", and therefore they had to wait on a 4th result. Since the 4th result wasn't returned by deadline, a new "result" was generated and sent out. When the 4th result at last was returned, it passed validation and all other results also passed validation. A very easy example explains how this is possible: An example-validator needs 3 results and uses error-limit 1. result1 = 50 result2 = 50 result3 = 52 Here one of the 3 results is outside the error-limit, meaning none passes validation. But when result4 = 51 is returned, all results sneaks inside the error-limit and all is validated. :) |
Don Erway Send message Joined: 18 May 99 Posts: 305 Credit: 471,946 RAC: 0 |
Hi Ingleside. Thanks for the clear explaination. But what if the 4th result, is 52? Would it ask for a 5th result? Does the validator really have no way to take the mean here? Should it? Thanks! Don |
Jos Ouddeken Send message Joined: 7 Apr 05 Posts: 18 Credit: 99,304 RAC: 0 |
Quote: Technews In addition, there are a great many result files in our upload directories that have no corresponding row in the database. These disassociated result files will never be deleted by the file deleter program. Such results can appear when a workunit had reached it's quorum number of returned results and is passed through validation, assimilation, file (both workunit and result) deletion and finally DB purging and *then* one or more results come in (perhaps they were slowed down by running intermittently on a laptop). The disassociated results are the bulk of what needs deleting. Unquote It seems to me, that if these disassociated results are the bulk of what needs deleting, why not change the upload procedure so that results which are NOT in the database simply cannot be uploaded ??? <img border="0" src="http://boinc.mundayweb.com/one/stats.php?userID=1691&prj=1&trans=off" /> <img src="http://mysite.wanadoo-members.co.uk/thefinalfrontear/babysm.JPG" /> |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
It seems to me, that if these disassociated results are the bulk of what needs deleting, why not change the upload procedure so that results which are NOT in the database simply cannot be uploaded ??? Thank you Jos Ouddeken this is the perfect example of a new user trying to be helpful. I'm not skilled enough to know if your idea has merit, but it certainly seems to be rendered with a good intent. Thank you tony |
ML1 Send message Joined: 25 Nov 01 Posts: 20147 Credit: 7,508,002 RAC: 20 |
It seems to me, that if these ... ??? LOL! That's one of the most beautifully subtle and gentle put-downs that I've seen on these forums for a good while. Nice one! I guess there must be a lot of eager naivety all round from many people wishing for this project to do well. In short: The presently large WVF queue is due to an old bug that has now been debugged. All should be fixed over the next week or two. Meanwhile, WU processing continues uninterrupted, and everyones' credits will catch up some time later. Happy crunchin', Regards, Martin See new freedom: Mageia Linux Take a look for yourself: Linux Format The Future is what We all make IT (GPLv3) |
Ingleside Send message Joined: 4 Feb 03 Posts: 1546 Credit: 15,832,022 RAC: 13 |
In the example-validator, you'll need a 5th result. As for taking the mean.... let's say you've got one signal at 1, another at 99, and a 3rd at 50... Now, the mean is 50, but would you really say this is the same signals? |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
That's one of the most beautifully subtle and gentle put-downs that I've seen on these forums for a good while. I certainly was NOT putting Jos down, He IS trying to be helpful. I can't help it if some other "Unnamed" individual notices the difference between helpful posts and the other kind. LOL |
Bill Barto Send message Joined: 28 Jun 99 Posts: 864 Credit: 58,712,313 RAC: 91 |
It seems to me, that if these disassociated results are the bulk of what needs deleting, why not change the upload procedure so that results which are NOT in the database simply cannot be uploaded ??? Or maybe uploaded but not added to the database. The user would get some kind of error message back. |
ML1 Send message Joined: 25 Nov 01 Posts: 20147 Credit: 7,508,002 RAC: 20 |
That's one of the most beautifully subtle and gentle put-downs that I've seen on these forums for a good while. Sorry, my "sarcasm/satire" suspicions were still switched on from Mr No Comment in a certain other thread. (Phew, that one can ramble general gibberish in great literary volumes enough to sink a library!...) As for the disassociated results fix suggestion... Mmmm, well... Shall we just say: "That is exactly the problem that is being worked on at the moment"? There are Dev comments elsewhere to say that a number of fixes are being considered for that problem. Very likely it is at the very centre of all their attention, and has been for the last few days. I certainly don't know the fine low level detail of the Boinc and Boinc-s@h protocols and database and filestore interactions. I would not presume to suggest how to fix that one. And the fix is not easy, and they likely do not wish to lose any results or cause silly upset with lost credits. Given a few hundred hours poring over the code and then perhaps... How many s@h crunchers does it take to replace a light bulb I wonder...? Regards, Martin See new freedom: Mageia Linux Take a look for yourself: Linux Format The Future is what We all make IT (GPLv3) |
Paul D. Buck Send message Joined: 19 Jul 00 Posts: 3898 Credit: 1,158,042 RAC: 0 |
Well, 2-3 to do the actual location of the problem, deciding how to fix it, by changing of the burned out bulb, repair the wiring, pay the electric bill, etc. 10-15 to point out that not only is candle light romantic, but the lights are on in other places and we can go there for the time being. Hoards to point out that BOINC is fatally flawed because it did not have redundent wiring, everlasting light bulbs, etc. and to insist that they are entitiled to life long light in this room without fail and that those that did not prevent this current disaster are incompetent and if they had only listend to the warnings they had made about the potential problmes we would not be in this predicament, and lastly the Classic light bulbs were good enough and should never have been considered for replacement by this obviously deficient design ... |
Astro Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
Well,.. Gee, Paul, that seems very analogous to the current Boinc/Seti posts were are seeing. You a funny man |
©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.