Posts by Ananas

Message boards : Number crunching : Server status = up but upload fails

( Message 7061 )
Posted 1623 days ago by Ananas

I did read that but the server came back with an current status page and unreachable services so there must be something going on besides the regular maintenance.

Message boards : Number crunching : Database purge ?

( Message 7051 )
Posted 1625 days ago by Ananas

I guess the database purge doesn't work as expected as it most likely uses the workunit table as the leading relation.

There are results though that are not attached to a workunit anymore, e.g. :

resultid=34595540 is an "orphan".

I have reported it somewhere else in this forum already but those results are still there.

p.s.: here is my old report about this problem.

Message boards : Number crunching : Server status = up but upload fails

( Message 7050 )
Posted 1625 days ago by Ananas

All servers running, status page is up to date but BOINC tells me there's no connect possible on uploads

Temporarily failed upload of ...: system connect

and on scheduler connects :

Scheduler request failed: Couldn't connect to server

An IP change? A firewall issue?

Message boards : Docking@Home Science : hi

( Message 7017 )
Posted 1663 days ago by Ananas

No hosts, no team, a "hello" that could be posted to any forum and a commercial URL?

Sounds not too much like a cruncher to me.

Message boards : Number crunching : Boinc funziona da solo? Il mio usa troppa CPU 100%, soluzione riduzione?

( Message 6998 )
Posted 1705 days ago by Ananas

You can limit the CPU usage in your global preferences (e.g. use less cores or limit the percentage)

BOINC uploads and reports the finished results on the next server contact - i.e. when it downloads more work or when the results are close to the deadline. No mailing required.

Message boards : Number crunching : Orphaned results in database

( Message 6986 )
Posted 1736 days ago by Ananas

*bump*

You should do a cleanup run, something like

"delete from result where result.workunitid not in (select distinct workunit.id from workunit)"

or

"delete from result where not exists (select distinct workunit.id from workunit where workunit.id = result.workunitid"

Message boards : Number crunching : Incorrect Function 1

( Message 6940 )
Posted 1748 days ago by Ananas

hmmmm ... the non-1hv*** results are getting rare, maybe it's time for another timeout. The last one had been caused by series having tons of "ERROR - Charmm exited with code 1." but no one has fixed it since then and no one seems to care about tons of crashing results at all.

Message boards : Number crunching : Incorrect Function 1

( Message 6939 )
Posted 1749 days ago by Ananas

Add 1hvj1htf and 1hvk1htf to the badlist

Message boards : Number crunching : Incorrect Function 1

( Message 6937 )
Posted 1749 days ago by Ananas

1hvi1htf is another buggy series, I'll abort all I get from this type.

10)

Message boards : Number crunching : Incorrect Function 1

( Message 6929 )
Posted 1753 days ago by Ananas

Out of the current batch of 1ohr1htf units I've had 7 errors out of 121 crunched so far and those were limited to 1 PC - a 1090T hex core. ...

Your box started working fine when you ran out of 1ohr1htf and received 1t7k1htf instead. I doubt that it is a voltage issue, it would have done that with the lower voltage too I bet. (You had way more than 7 errors btw. and all only in certain series)

1dif1htf (those that caused problems for me) fail on your box too and on all other hosts I checked.

Mine is a Xeon L5520, standard voltages and frequencies.

@project : Again ... as this "ERROR - Charmm exited with code 1." is a program controlled exit, you should "exit 0" and set a flag that the result ran into a condition where further processing doesn't make sense for scientific reasons. On BOINC-side it should be successfull. Compare it to a prime project - there the numbers that turn out to be no prime do not exit with an error either.

Next 10 posts