I think this has happened to milkyway@home more than a couple times -- is there any way to have a more robust way to check the number of workunits ready to send so the database doesn't get overloaded if things go wrong?
--Travis On Oct 11, 2011, at 11:08 AM, Richard Haselgrove wrote: > It certainly happened at SETI a few days ago when the transitioner went > offline (the server hosting the transitioner process crashed). 'Results ready > to send' peaked at about 1,750,000 - the usual limit keeps it between 200,000 > / 250,000 - from which we deduce that the splitters (their workunit > generators) carried on running while the displayed figures weren't being > updated. In their case, the database was comparatively lightly loaded at the > time, and some swift homework by a staff member prevented it getting out of > hand, but it was a near miss. > >> Is it possible for the work generator to go a bit haywire? If we're using a >> policy which checks to see the number of unsent workunits, if one of the >> feeder queries gets hung up, is it possible for the work generator to keep >> thinking that there aren't enough unsent workunits because the feeder hasn't >> updated things and then repeatedly generate more workunits which could flood >> the database? >> >> I'm wondering if this is a possibility because I think it's what might have >> happened with the most recent MilkyWay@Home crash... >> >> thanks, >> --Travis > > --------------------------------------------------------------------------- Travis Desell, Assistant Professor University of North Dakota - Dept. of Computer Science [email protected] - cell: 518-867-1054 Streibel Hall Room 220 - office: 777-701-3477 3950 Campus Road Stop 9015 Grand Forks, North Dakota 52802-9015 Homepage ( http://people.cs.und.edu/~tdesell/ ) MilkyWay@Home ( http://milkyway.cs.rpi.edu/ ) DNA@Home ( http://dnahome.cs.rpi.edu/ ) Worldwide Computing Laboratory ( http://wcl.cs.rpi.edu/ ) ---------------------------------------------------------------------------- _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
