The default is 25%, but yes, that might well lead to a situation in which tasks keep starting and stopping, especially on a small host like that. Practically any normal usage will have the load that high and if the rig is not left to run when nobody is working at it, there may not be long enough uninterrupted stretches. Anyway, as Richard pointed out, this is not only about practical defaults and getting enough information via stderr, but mainly about user PR. It won't help if you promote the platform (e.g. the Android launch) , but offer a substandard product
2014-03-18 16:55 GMT+01:00 Raistmer the Sorcerer <[email protected]>: > Perhaps one should blame suspend on >20% load default setting. > too many suspends lead to task computational error you describe. > Not wise defaults. > > > Tue, 18 Mar 2014 15:05:22 +0000 (GMT) от Richard Haselgrove < > [email protected]>: > >Looking around at SETI, trying to analyse how well the server was > performing while the status page was dysfunctional, I came across this user. > > > >http://setiathome.berkeley.edu/show_host_detail.php?hostid=7220378 > > > > > >He (and his - somewhat elderly - computer) have been members for about 3 > weeks. He's got a little credit, both at SETI and at Cosmology, so the > computer seems to be in working order. > > > >But now, every single task - at both projects, today - is failing > with Exit status-226 (0xffffffffffffff1e) ERR_TOO_MANY_EXITS (oh, and a few > download errors). > > > >He seems to be using what was the recommended BOINC at the time > (v7.2.39), stock science applications, and I'd be willing to bet default > BOINC computing settings. BOINC has just *got* to work for volunteers like > that, out of the box - otherwise there's no point. > > > >So, as a reasonably experienced "reader of the runes", what can I deduce > from the error messages presented that I might possibly pass on in a > helpful private message? Er, as it happens, nothing. I'm stuck. Even task > http://www.cosmologyathome.org/result.php?resultid=28640358 - which ran > for nearly two hours - presents nothing useful in stderr_txt. > > > >If we're going to seal the black box, so that no information leaks out to > the target "volunteer spare cycles" user, then the onus is on us to make > sure it works. First time, every time. > > > >Otherwise it's no surprise the number of volunteers goes down, year after > year. > >_______________________________________________ > >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. > > _______________________________________________ > 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. > _______________________________________________ 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.
