log in
1) Message boards : Number crunching : Strange granting values (Message 762)
Posted 21 May 2008 by Profile MikeMarsUK

I doubt that the scientists would reject a result because of a quorum issue. Certainly on the various CPDN projects the quorum is entirely irrelevant from their viewpoint.
2) Message boards : Number crunching : Validate errors part 2 (Message 740)
Posted 12 May 2008 by Profile MikeMarsUK

I\'ve got a 512MB XP/SP2 laptop which seems to be working fine (95% success or thereabouts). But notice that does nothing other than run Boinc on this project - other projects are disabled (hence their workunits will not be taking up memory), and I don\'t use the laptop for anything else. I also try to strip down Windows as much as possible whenever I get hold of a new machine.


http://mindmodeling.org/beta/show_host_detail.php?hostid=2929
3) Questions and Answers : Windows : Computation Error (Message 710)
Posted 6 May 2008 by Profile MikeMarsUK

You need Boinc version 5.5.1 or higher, otherwise the workunits will crash.

I\'ve suggested (on a different thread) that this version limit be set explicitly, although I haven\'t seen a response.

4) Questions and Answers : Windows : Memory Leak (Message 709)
Posted 6 May 2008 by Profile MikeMarsUK
All mine have that message, and I get credit for 90% of them. I think you may have a corrupted lisp installation.

Try this:

* Set Boinc to only use one Core

* Detach and then reattach to the project (so it deletes the project folders)

* Run a single workunit to completion. This step should re-install the LISP environment.

* Once it has finished, set up Boinc to use the full amount of Cores
5) Questions and Answers : Windows : Memory Leak (Message 704)
Posted 4 May 2008 by Profile MikeMarsUK

Pretty small memory leak (just 5K in total), particularly since it will be returned to the system as soon as the workunit finishes. I wouldn\'t worry about it too much.
6) Questions and Answers : Unix/Linux : 44 (of 44) units with errors (Message 681)
Posted 26 Apr 2008 by Profile MikeMarsUK

Kathryn has made a good point - have you installed the Ubuntu 32-bit package support module? (It\'s nothing to do with Boinc, but is required to run a lot of software).

The following instructions may help with the 32-bit package support installation: http://www.climateprediction.net/board/viewtopic.php?t=5896 (look for the \'Linux\' section).

If you see \'running\' with the CPU time not increasing, then it is probably installing the LISP environment. The first WU of each type needs to do this, and if interrupted you may end up with a corrupted Lisp environment, in which case you need to delete the appropriate folder, and start again.
7) Message boards : Number crunching : WUs remain simply hang ...! (Message 665)
Posted 22 Apr 2008 by Profile MikeMarsUK

Your computers are hidden, so we can\'t see if there are any known problems there (i.e., Win64 for example).

I\'ve run about 20 so far and they\'ve all worked fine for me...
8) Message boards : Number crunching : Strange granting values (Message 661)
Posted 22 Apr 2008 by Profile MikeMarsUK

I\'ve had a couple like these - successful outcome, 90 to 120 minutes of processing, but zero grant:

http://mindmodeling.org/beta/result.php?resultid=1251619

I think it\'s because everyone else running the same workunit crashed out.


Ideally the credit should ignore the benchmark entirely, the benchmarks are a huge can of worms. Are there any metrics from ACT_R which can be used instead, preferably to predict the amount of processing required? How about leaving the credit \'pending\' until a few hundred results are in, and then using that as a basis?

It may also be interesting to maintain a ratio between credit requested, and credit granted, on a host-by-host basis. This could be used as a multiplier. (A poor solution, but still better than looking directly at the benchmarks).

9) Message boards : Number crunching : Error code -185 / Couldn\'t start or resume: 2 (Message 660)
Posted 20 Apr 2008 by Profile MikeMarsUK

I found some more examples of this error at this project, but none of the affected users have posted a query at the forum:

http://www.google.co.uk/search?q=+%22Couldn%27t+start+or+resume:+2%22&hl=en&start=10&sa=N&filter=0

I also suspect it may be due to the Boinc version. All the affected users have 5.4.x, and the google results contained another project suffering from this error where it was stated that the minimum Boinc version was 5.5.1.

If it is confirmed that there is a minimum Boinc version, could this be set in the project options, so that an appropriate message appears when attempting to download new work? I know an option exists to do that, but I don\'t recall what it is.

-- Edit:

The following page mentions v5.5 again, and also has links to where the minimum version can be set.
http://boinc.berkeley.edu/trac/wiki/WrapperApp
10) Message boards : Number crunching : Error code -185 / Couldn\'t start or resume: 2 (Message 659)
Posted 20 Apr 2008 by Profile MikeMarsUK
...
Curiously the most recent task is sitting there \'ready to run\' with Boinc having preempted my other project. I notice on some of the other threads that it is recommended to leave it alone (don\'t abort) in that status, so I\'ll do that....


After a couple of hours I forced the stuck task to run (by suspending and reenabling the preempted task from the other project), but it crashed in the same way as the others.
11) Message boards : Number crunching : Error code -185 / Couldn\'t start or resume: 2 (Message 658)
Posted 20 Apr 2008 by Profile MikeMarsUK
I\'m trying to attach a laptop to the project:

http://mindmodeling.org/beta/workunit.php?wuid=380653

But it\'s killing tasks instantly with an Exit status -185. The following is in the error text:
<core_client_version>5.4.11</core_client_version> <message> Couldn\'t start or resume: 2 </message>


The Boinc message log says:

20/04/2008 09:55:25|MindModeling@Beta|Can\'t copy projects/MindModeling.org_beta/PVTNM_noattend_large-3-0.550_0.0275_0.775_2.275_1208585342_3_0 to slots/1/out
20/04/2008 09:55:25|MindModeling@Beta|Unrecoverable error for result PVTNM_noattend_large-3-0.550_0.0275_0.775_2.275_1208585342_3 (Couldn\'t start or resume: 2)
20/04/2008 09:55:25|MindModeling@Beta|Deferring scheduler requests for 1 minutes and 0 seconds
20/04/2008 09:55:25||Rescheduling CPU: start failed
20/04/2008 09:55:25|MindModeling@Beta|Unexpected state 7 for task PVTNM_noattend_large-3-0.550_0.0275_0.775_2.275_1208585342_3
20/04/2008 09:55:26|MindModeling@Beta|Computation for task PVTNM_noattend_large-3-0.550_0.0275_0.775_2.275_1208585342_3 finished
20/04/2008 09:55:27||Rescheduling CPU: result suspended, resumed or aborted by user


Any ideas what the resolution for this is? The laptop is running Boinc 5.4.11, XP SP2, and has 512MB ram, single core (Pentium-M).

-- Edit:

Curiously the most recent task is sitting there \'ready to run\' with Boinc having preempted my other project. I notice on some of the other threads that it is recommended to leave it alone (don\'t abort) in that status, so I\'ll do that....




Main page · Your account · Message boards


Copyright © 2020 MindModeling.org