Message boards :
Number crunching :
Accurate errors?
Message board moderation
Author | Message |
---|---|
Prognatus Send message Joined: 6 Jul 99 Posts: 1600 Credit: 391,546 RAC: 0 |
When looking at one of my returned results...: http://setiweb.ssl.berkeley.edu/workunit.php?wuid=10116868 ..I wonder why I'm the only one (of 8) who actually could process this particular WU. Well, one other did process it, but claimed no credits. (Also strange that he/she didn't use any time to process this WU!?!!) However, it seems somehow unlikely that all the other 6 should get the same error: "Client error". I know there could be lots of possible reasons for error on the client side, but how likely is it that so many should get it on the same WU? And why did one computer use zero second on the same WU? What's happening here? My question is this: are these errors accurate? I mean, do they always tell the right reason? I'm beginning to doubt it... BTW, here's the list of possible outcomes, for reference: http://setiweb.ssl.berkeley.edu/explain_state.php?field=result_outcome |
Ingleside Send message Joined: 4 Feb 03 Posts: 1546 Credit: 15,832,022 RAC: 13 |
> When looking at one of my returned results...: > > http://setiweb.ssl.berkeley.edu/workunit.php?wuid=10116868 > 7 of the results is download-errors, the "success" with zero cpu-time is also download-error, this shows up in stderror-out, so should have been marked as "download-error" instead of "success". 6 of the clients with download-errors is running v4.13, a client that was replaced with v4.19 some time ago due to a bug in downloading, there if fails download it error-out instead of re-trying... With high server-load it's a higher chance for download-problems, so v4.13-clients failing to download isn't unexpected. The last is with v4.53, an alpha-client that isn't even feature-complete, had the download-bug, and to add on top a fatal upload-bug... so even if had managed to download successfully it would have been very unlikely managed to upload correctly due to all the outages, and the high server-load when was up. |
JAF Send message Joined: 9 Aug 00 Posts: 289 Credit: 168,721 RAC: 0 |
> When looking at one of my returned results...: > > http://setiweb.ssl.berkeley.edu/workunit.php?wuid=10116868 > > ..I wonder why I'm the only one (of 8) who actually could process this > particular WU. Well, one other did process it, but claimed no credits. (Also > strange that he/she didn't use any time to process this WU!?!!) However, it > seems somehow unlikely that all the other 6 should get the same error: "Client > error". I know there could be lots of possible reasons for error on the > client side, but how likely is it that so many should get it on the same WU? > And why did one computer use zero second on the same WU? What's happening > here? > > My question is this: are these errors accurate? I mean, do they always > tell the right reason? I'm beginning to doubt it... > I've put forth this message before and haven't gotten much response, but I'd like to see a self test work unit for each project -- maybe initiated through a BoincMGR menu item. If one is getting errors, select self test and download and crunch the reference WU. Upload and see if it errors or not. This might help over-clockers and cut down on the message traffic (at least we could respond with "did you run the self test?") Seems like this would check the whole process: download, crunch, upload, and report. Bjorn, in this case I suspect it was just bad luck on when you crunched the WU and when there were project problems causing download/upload/reporting errors. <img src='http://www.boincsynergy.com/images/stats/comb-912.jpg'> |
Prognatus Send message Joined: 6 Jul 99 Posts: 1600 Credit: 391,546 RAC: 0 |
OK, I see that this had explainable reasons and bad luck - however strange it looked first. It makes you wonder though... if the people still running 4.13 knows about how bad it can be, since they haven't upgraded to 4.19 yet. On the philosophical side: If 4.19 corrected a bug in 4.13 about download, then why are these categorized as "Client error"? It may be the client side who generates the error, but since it's because of a bug in the SETI software, it's hardly linked to the client itself, but SETI. So, the blame should be on the server side. |
Ingleside Send message Joined: 4 Feb 03 Posts: 1546 Credit: 15,832,022 RAC: 13 |
> On the philosophical side: If 4.19 corrected a bug in 4.13 about download, > then why are these categorized as "Client error"? It may be the client side > who generates the error, but since it's because of a bug in the SETI software, > it's hardly linked to the client itself, but SETI. So, the blame should be on > the server side. > The bug was in the BOINC core client, not in the seti-application, and not in the seti-servers. |
©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.