log in
1) Message boards : Number crunching : Reported back too late to validate? (Message 1063)
Posted 22 Oct 2008 by Dagorath
Yep. IR = 5 and Min.Q = 5. What a disgusting waste of donated computing power. I've added this project to my List of Welfare Bum Projects and have detached my hosts. They will not be reattached.

Do it right, Jack, or buy the computing power yourself!!!

2) Questions and Answers : Windows : Lack of work ? (Message 1045)
Posted 30 Sep 2008 by Dagorath
Just wondering if anything is wrong ? Still not had any work for about a week now.


If you're not getting any work then one of the first things you should check is the Server Status. To see the Server Status go to the home page and see near the top on the right hand side where it says "Click here for server status". That takes you to the Server Status page. Almost every BOINC project has link to their Server Status somewhere on their home page.

On the Server Status Page, check the line that says "Results ready to send". If that says 0 then the project has no work available.

Over on the left side of the page you'll see the names of various programs that run on the server. The scheduler, feeder, transitioner and upload/download server are involved in producing, scheduling and sending work to the hosts attached to the project. If any of those 3 are down, especially the scheduler and upload/download server, then you might not get any work even if "Results ready to send" says work is available.

3) Message boards : Number crunching : Reported back too late to validate? (Message 999)
Posted 13 Sep 2008 by Dagorath
The latest bunch of WU (and the template used to create new ones) now reflect a
1 - initial replication
10 - Max Error

Plus Validation Errors have become MUCH less common since 3.4+ If something goes wrong - things are identified on the client and a "Computation Error" occurs. The "Why" did it die is now a bit easier to answer because an error code is returned to the server. These codes can be found here

Thanks,
--Jack


Why can't all project admins be as reasonable and quick to respond with real solutions.

4) Message boards : Number crunching : Validation Errors (Message 977)
Posted 9 Sep 2008 by Dagorath
So, just kind of new to this project - just started crunching. So forgive me if these things have already been discussed. I assume they have been.


It's being discussed in this thread already.

@Bailiff: whack his RAC. 10K credits for first offense, 20K if it happens again. Forgiveness, YEESH! We're running a BOINC project here not a confessional.

5) Message boards : Number crunching : Reported back too late to validate? (Message 976)
Posted 9 Sep 2008 by Dagorath
I presume this means that we can expect 66% of our crunching to go "validate error" and get 0.00 credits for our time and effort?

I believe that if we have crunched the wu in good faith, and no cheating of any kind has been done, that good faith should be rewarded by the project, whether it be alpha, beta, or full production...

or am I missing something here?


You're confusing 2 separate issues. The 2 issues are related but the 66% figure is true in only 1 of those issues and it's not the validate error issue. Maybe I can separate and clarify the 2 issues for you.

You will get the validate error when the chain of events occurs the way BobCat13 describes in the last paragraph of message #960 in this thread. Read that paragraph carefully. It is complicated but BobCat13 has said it in as simple terms as possible, at least I don't think I can say it better than he did so I'm not going to try. Yah, lazy me :)

How often will the chain of events occur the way BobCat13 describes? I don't know. It may be 66% of the time or it may be far less. BobCat13 is saying that the chain of events can be avoided (in other words the validate errors can be avoided) by reducing the initial replication to the quorum number which is 1.

The 66% I mentioned in my message #970 is the amount of wasted computing time that will occur if this project has an initial replication of 3 but needs only 1 successful result for each work unit to make their final analysis a valid analysis. That is a different topic. I brought it up only to point out to Jack that there may be a second reason to follow BobCat13's advice to reduce the initial replication to the quorum number.

6) Message boards : Number crunching : Reported back too late to validate? (Message 970)
Posted 7 Sep 2008 by Dagorath
And another thing, if you are going to send out 3 WU's in the initial replication, then if all three WU's come back in time and have valid data they should all get credit.

I didn't get credit on this becaus I reported back 4 minutes after someone else, and I crunched it quicker.


You are correct - that would only be fair.
Until I can rewrite the validator - I re-ran a script to fix credit; this method is far from ideal because it does not affect RAC only total credit counts.

I will see what I can do,
--Jack


Jack, the fact that the server declares the first result that returns a Success to be the Canonical Result before any other results are returned indicates that you don't really need the other 2 results. That means 66% of the CPU cycles donated to MindModeling are wasted. If you were the only project and there was an oversupply of CPU cycles then I suppose it wouldn't matter much. Unfortunately, there are many other projects that need our spare CPU time too. Every cycle that goes to waste here is a cycle that could have been doing something useful at some other worthy project. If you're doing it just to speed up the return of results then ask yourself what the other projects should do to overcome the slowdown you are causing by wasting cycles. Maybe they should make their initial replication 5 when they only need 2 results? Jack, can we really get all the work done faster by doing work we don't need to do?

7) Message boards : Number crunching : What happened? (Message 880)
Posted 23 Jul 2008 by Dagorath
Admins et al,

Instead of us repeatedly checking the board is there a project status that I can check against without browsing the board for updates?

tia,
Impatient Roland


If you're browsing the message boards for updates then you're doing it wrong and working harder than you need too. In that case you'll be glad to know about the Server Status page.

Almost every BOINC project has a Server Status page and a link to that page somewhere on their home page. This project has one too.

If you're really lazy (like me) and consider checking Server Status to be too much work then there are scripts around that will periodically request the xml file from which the Server Status page is built and, if work is available, click the update button for you while you sip a cold one at the 19th hole. Can't win the scramble without it ;)



Or, you could just leave the project always running and always accepting work. Then, when work shows up on the servers your machine downloads it


That strategy doesn't always work. If your 'puter is requesting work from a project that doesn't have work, the delays between requests grow longer and longer. Eventually the delay will grow to several hours. When the project starts sending work other crunchers sometimes take it all before your computer sends another work request. Later, when your computer requests work, there is none left. That bothers various people for various reasons, believe it or not.

Of course, I realize that many BOINCer's like to micro-manage.


Many BOINCers have neither the time nor the inclination to micromanage. BOINC automatically does everything they want it to do so no need to micromanage. Others like competing to see who can get the most work out of a project where work is scarce and that requires a little micromanaging. Whatever blows their hair back, I suppose.

8) Message boards : Number crunching : What happened? (Message 878)
Posted 23 Jul 2008 by Dagorath
Admins et al,

Instead of us repeatedly checking the board is there a project status that I can check against without browsing the board for updates?

tia,
Impatient Roland


If you're browsing the message boards for updates then you're doing it wrong and working harder than you need too. In that case you'll be glad to know about the Server Status page.

Almost every BOINC project has a Server Status page and a link to that page somewhere on their home page. This project has one too.

If you're really lazy (like me) and consider checking Server Status to be too much work then there are scripts around that will periodically request the xml file from which the Server Status page is built and, if work is available, click the update button for you while you sip a cold one at the 19th hole. Can't win the scramble without it ;)

9) Message boards : Number crunching : Strange granting values (Message 622)
Posted 9 Apr 2008 by Dagorath
This will reduce quorum to 1,normalize credit give by MindModeling as compared to other volunteer projects, remove cheating, and standardized credit given to Mac, Linux, and Windows volunteers.

This will take a little time to implement, but I think it is a better solution in the end.

Thanks for your help,
Jack


A credit scheme as grand as that deserves a grand acronym name. How about SNRQNCOSS (pronounced snerknicoss)?

SETI
Normalised
Reduced
Quorum
Non
Cheatable
Operating
System
Standardised

10) Message boards : Number crunching : Strange granting values (Message 617)
Posted 9 Apr 2008 by Dagorath
Good work... but... you\'ve opened the door to credit cheaters who manipulate their benchmarks upward which makes their host submit wildly inflated credit claims. The stock code isn\'t fair for the reasons noted in posts earlier in this thread but it has the advantage that it tosses out cheater claims. Your fix keeps cheater claims and awards both crunchers far more than they earned.


No. It awards cheaters the average of their claim and the wingman\'s claim.


Sorry I wasn\'t explicit but I thought readers would be able to fill in what I left out. Where I said \"Your fix keeps cheater claims\" I meant the cheater claim is kept and used in the averge. Anyway, both members of the quorum are awarded more than they earn.

Yes, it can be inflated. But that beats getting awarded crap because a wingman\'s machine claims crap.


And if neither your or your wingman are cheating then you\'re earning crap compard to what cheaters are earning. That\'s the downside to Jack\'s fix and though it doesn\'t seem to bother you it bothers a lot of other crunchers.

@Jack, consider abandoning the benchmark based credit system and implement fixed credits. It\'s the only way to eliminate the cheating.


The intermediate mechanism I am using to help cheating is that I am validating the results on return. However it is the case that good data returned with bad claimed credit slips through.

Fixed credit is difficult because the different models run on MM actually take different amounts of time and different parameter combinations run on the models also influence runtime.


I understand so no presure from me. Thanks for listening and responding. You\'ve done a great job so far. Keep up the good work.

11) Message boards : Number crunching : Strange granting values (Message 612)
Posted 9 Apr 2008 by Dagorath
Good work... but... you\'ve opened the door to credit cheaters who manipulate their benchmarks upward which makes their host submit wildly inflated credit claims. The stock code isn\'t fair for the reasons noted in posts earlier in this thread but it has the advantage that it tosses out cheater claims. Your fix keeps cheater claims and awards both crunchers far more than they earned.


No. It awards cheaters the average of their claim and the wingman\'s claim.


Sorry I wasn\'t explicit but I thought readers would be able to fill in what I left out. Where I said \"Your fix keeps cheater claims\" I meant the cheater claim is kept and used in the averge. Anyway, both members of the quorum are awarded more than they earn.

Yes, it can be inflated. But that beats getting awarded crap because a wingman\'s machine claims crap.


And if neither your or your wingman are cheating then you\'re earning crap compard to what cheaters are earning. That\'s the downside to Jack\'s fix and though it doesn\'t seem to bother you it bothers a lot of other crunchers.

@Jack, consider abandoning the benchmark based credit system and implement fixed credits. It\'s the only way to eliminate the cheating.
12) Message boards : Number crunching : Strange granting values (Message 609)
Posted 9 Apr 2008 by Dagorath
I have rewritten the code.

It is now returning a strict average.
--Jack


Good work... but... you\'ve opened the door to credit cheaters who manipulate their benchmarks upward which makes their host submit wildly inflated credit claims. The stock code isn\'t fair for the reasons noted in posts earlier in this thread but it has the advantage that it tosses out cheater claims. Your fix keeps cheater claims and awards both crunchers far more than they earned.

13) Message boards : Number crunching : RESOLVED - WUs Completed with sucess Errors. (Message 443)
Posted 7 Mar 2008 by Dagorath
I don\'t know if it\'s just me but all WUs returned Completed with Sucess are getting these results \"Workunit error - check skipped\" since the new Version came available.

Do I have to reset project for all my host to get new Version 3.0 instead of 2.61 to stop WUs from being skipped?

Thanks

Bill

Bill,

I checked your results for host 1086 and those workunits have been cancelled by the server, so if you have no version 3.00 tasks just reset the project and get some v3.00 work.


BobCat13,

Thanks. I already reset that one because of Work Unit Errors. I didn\'t know whether to do them all or not. Today is the first time I\'ve noticed the error.

Thanks Again and happy Crunching,

Bill


When did you reset? If you reset before v3 started coming down and you took down more v2.61...
14) Message boards : Number crunching : Linux app not running (Message 438)
Posted 6 Mar 2008 by Dagorath
Version 3.0 has extra size safeguards.
How is it working now for you?

--Jack


I have 2 machines with identical hardware, running Fedora 5, 875 MB RAM. Version 2 ran OK on 1 of those machines but on the other about 50% of the WUs ended with Compute Error. So it seems with version 2, 875 MB RAM was borderline... sometimes enough, sometimes NOT enough.

Now, version 3 seems to be running fine on both those machines. The extra size safeguards seem to be working very well. Also, top says mm_sbcl is getting about 98% of CPU so the wrapper app seems to be taking negligible CPU time.

I also have a machine with 495 MB RAM, running Fedora 8. The first WU got to about 95% complete then it crashed BOINC core (client). I\'ve never seen a WU do that. I rebooted, restarted BOINC and it crashed again, stdoutdae says it can\'t rename some file, sorry, don\'t remember the filename.

WUs also crash on that machine when it\'s booted on Win XP so it seems to me 495 MB RAM just isn\'t enough for version 3.
15) Questions and Answers : Windows : 64-bit BOINC client / 32-bit work. (Message 364)
Posted 13 Feb 2008 by Dagorath
OK, scratch machines with 512K or less RAM.

One of my other machines has been crunching MM units sucessfully on 64 bit Linux. It has 1 GB RAM though the video steals a bit and leaves it with 895 MB. I booted that one on 32 bit XP and tested with various versions of sbcl from the link you give above as well as 1.0.9. I conclude that 895 available RAM is not enough, the reasons are...

The best results were obtained with sbcl 1.0.9. About 50% of the work units ran for about 20 minutes, showed regular %complete increments and ended with success. A few of those have already verified against results from wingmen. The other 50% crashed within 1 minute.

I had to kill explorer.exe (to save RAM) and run boinc.exe in a DOS box to get 50% successes otherwise the work crashed.

Also tried sbcl 1.0.14.7, 1.0.14.22 and 1.0.14.14 hoping one of those implement dynamic memory allocation but it seems they don\'t. All work units finished within 1 minute (wall clock time) and claimed < 1 second CPU time but reported success. I think the success status is erroneous but may be wrong. Here is one such result. Note that work unit was crunched twice on my machine (host 1208) and now has status Done and a canonical result declared. I bet neither of those results would verify against a result from a machine with adequate RAM. Hope you\'re somehow screening canonical results because that one is likely BS.

Perhaps I installed the newer versions of sbcl incorrectly? They are distributed as a .msi file. I simply uninstalled previous version via Add/Remove Software, let the installer do its thing then copied the sbcl.exe and sbcl.core into ...boinc/projects/mindmodeling.org_beta/sbcl1.0.9.

16) Questions and Answers : Windows : 64-bit BOINC client / 32-bit work. (Message 362)
Posted 13 Feb 2008 by Dagorath
And us poor 32 bit XP users too. I have 3 XP 32 bit machines that crash every work unit. All have 512K ram and work fine at other projects.

It seems to be a problem with sbcl. I\'ve played with sbcl 1.0.09 and 1.0.13, running it in a DOS box with sbcl --core sbcl.core and found that it runs about 50% of the time (I get the \"death kitty\" message) and gives the \"unable to find enough memory message\" or something like that the other 50%. I\'ve killed all other unnecessary processes including anti-virus and set the page file to minimum 2 Gig and max 4 Gig. I\'ve played with the settings on Control Panel->System->Advanced->Performance Settings->Advanced and tried all combinations of options under Processor Scheduling and Memory Usage but still sbcl crashes 50% of the time and work units crash 100%.

Tonight I detached this machine, made sure .../projects/mindmodeling.org_beta got deleted to ensure no leftovers, then reattached and crunched 3 work units, task IDs are 608816, 608784 and 608747, I won\'t take any new work on that machine so they should be near the top of this page. The logged stderr out for all 3 is:

<core_client_version>5.10.40</core_client_version>
<![CDATA[
<message>
Incorrect function. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
app error running lisp: 0x1

</stderr_txt>
]]>

My stdoutdae.txt shows the following for those 3:

12-Feb-2008 22:32:09 [---] Fetching configuration file from http://MindModeling.org/beta/get_project_config.php
12-Feb-2008 22:32:37 [MindModeling@Beta] Master file download succeeded
12-Feb-2008 22:32:43 [MindModeling@Beta] Sending scheduler request: Project initialization. Requesting 1 seconds of work, reporting 0 completed tasks
12-Feb-2008 22:32:48 [MindModeling@Beta] Scheduler request succeeded: got 1 new tasks
12-Feb-2008 22:32:50 [MindModeling@Beta] Started download of ACTR_2.61_windows_intelx86.exe
12-Feb-2008 22:32:50 [MindModeling@Beta] Started download of watchdog.exe
12-Feb-2008 22:32:53 [MindModeling@Beta] Finished download of watchdog.exe
12-Feb-2008 22:32:53 [MindModeling@Beta] Started download of ACTR_graphics_2.61_windows_intelx86.exe
12-Feb-2008 22:32:55 [MindModeling@Beta] Finished download of ACTR_2.61_windows_intelx86.exe
12-Feb-2008 22:32:55 [MindModeling@Beta] Started download of Helvetica.txf
12-Feb-2008 22:32:57 [MindModeling@Beta] Finished download of Helvetica.txf
12-Feb-2008 22:32:57 [MindModeling@Beta] Started download of sbcl1.0.9.exe
12-Feb-2008 22:33:21 [MindModeling@Beta] Finished download of ACTR_graphics_2.61_windows_intelx86.exe
12-Feb-2008 22:33:21 [MindModeling@Beta] Started download of actr6v1.2r492_oldutility--sbcl1.0.9.tar.bz2
12-Feb-2008 22:33:28 [MindModeling@Beta] Finished download of sbcl1.0.9.exe
12-Feb-2008 22:33:28 [MindModeling@Beta] Started download of actr6v1.2r492_oldutility--sbcl1.0.9.exe
12-Feb-2008 22:33:30 [MindModeling@Beta] Finished download of actr6v1.2r492_oldutility--sbcl1.0.9.tar.bz2
12-Feb-2008 22:33:30 [MindModeling@Beta] Started download of PRP_Test4-jobinfo.xml
12-Feb-2008 22:33:31 [MindModeling@Beta] Finished download of PRP_Test4-jobinfo.xml
12-Feb-2008 22:33:31 [MindModeling@Beta] Started download of afrl-prp-gln-01n-use-fixed_g-ut-preds-boinc.lisp
12-Feb-2008 22:33:32 [MindModeling@Beta] Finished download of afrl-prp-gln-01n-use-fixed_g-ut-preds-boinc.lisp
12-Feb-2008 22:33:32 [MindModeling@Beta] Started download of PRP_Test4-2-0.26_1.75_3.32_0.017_1202237557
12-Feb-2008 22:33:33 [MindModeling@Beta] Finished download of PRP_Test4-2-0.26_1.75_3.32_0.017_1202237557
12-Feb-2008 22:33:33 [MindModeling@Beta] Started download of mm_ProjectIcon_BETA_01.png
12-Feb-2008 22:33:34 [MindModeling@Beta] Finished download of mm_ProjectIcon_BETA_01.png
12-Feb-2008 22:33:34 [MindModeling@Beta] Started download of activation.png
12-Feb-2008 22:33:35 [MindModeling@Beta] Finished download of activation.png
12-Feb-2008 22:33:35 [MindModeling@Beta] Started download of sideview.png
12-Feb-2008 22:33:36 [MindModeling@Beta] Finished download of sideview.png
12-Feb-2008 22:33:40 [MindModeling@Beta] Finished download of actr6v1.2r492_oldutility--sbcl1.0.9.exe

12-Feb-2008 22:33:41 [MindModeling@Beta] Starting PRP_Test4-2-0.26_1.75_3.32_0.017_1202237557_0
12-Feb-2008 22:33:41 [MindModeling@Beta] Starting task PRP_Test4-2-0.26_1.75_3.32_0.017_1202237557_0 using ACTR version 261
12-Feb-2008 22:34:13 [MindModeling@Beta] [error] Can\'t rename output file PRP_Test4-2-0.26_1.75_3.32_0.017_1202237557_0_0
12-Feb-2008 22:34:13 [MindModeling@Beta] Computation for task PRP_Test4-2-0.26_1.75_3.32_0.017_1202237557_0 finished
12-Feb-2008 22:34:13 [MindModeling@Beta] Output file PRP_Test4-2-0.26_1.75_3.32_0.017_1202237557_0_0 for task PRP_Test4-2-0.26_1.75_3.32_0.017_1202237557_0 absent

12-Feb-2008 22:34:13 [ABC@home] Resuming task abc_wu_3996258909000_2805000_1 using abc-finder version 103
12-Feb-2008 22:35:08 [MindModeling@Beta] Sending scheduler request: To fetch work. Requesting 706 seconds of work, reporting 1 completed tasks
12-Feb-2008 22:35:13 [MindModeling@Beta] Scheduler request succeeded: got 1 new tasks

12-Feb-2008 22:35:21 [MindModeling@Beta] [error] Couldn\'t delete file projects/MindModeling.org_beta/PRP_Test4-2-0.26_1.75_3.32_0.017_1202237557

12-Feb-2008 22:35:21 [MindModeling@Beta] Started download of PRP_Test4-2-0.26_1.81_2.56_0.003_1202237567
12-Feb-2008 22:35:22 [MindModeling@Beta] Finished download of PRP_Test4-2-0.26_1.81_2.56_0.003_1202237567
12-Feb-2008 22:35:23 [MindModeling@Beta] Starting PRP_Test4-2-0.26_1.81_2.56_0.003_1202237567_1
12-Feb-2008 22:35:23 [MindModeling@Beta] Starting task PRP_Test4-2-0.26_1.81_2.56_0.003_1202237567_1 using ACTR version 261
12-Feb-2008 22:35:37 [MindModeling@Beta] Computation for task PRP_Test4-2-0.26_1.81_2.56_0.003_1202237567_1 finished

12-Feb-2008 22:35:37 [ABC@home] Resuming task abc_wu_3996258909000_2805000_1 using abc-finder version 103
12-Feb-2008 22:35:38 [MindModeling@Beta] Started upload of PRP_Test4-2-0.26_1.81_2.56_0.003_1202237567_1_0
12-Feb-2008 22:35:40 [MindModeling@Beta] Finished upload of PRP_Test4-2-0.26_1.81_2.56_0.003_1202237567_1_0
12-Feb-2008 22:36:38 [MindModeling@Beta] Sending scheduler request: To fetch work. Requesting 706 seconds of work, reporting 1 completed tasks
12-Feb-2008 22:36:43 [MindModeling@Beta] Scheduler request succeeded: got 1 new tasks
12-Feb-2008 22:36:45 [MindModeling@Beta] Started download of PRP_Test4-2-0.26_1.81_1.91_0.022_1202237562
12-Feb-2008 22:36:47 [MindModeling@Beta] Finished download of PRP_Test4-2-0.26_1.81_1.91_0.022_1202237562

12-Feb-2008 22:36:48 [MindModeling@Beta] Starting PRP_Test4-2-0.26_1.81_1.91_0.022_1202237562_1
12-Feb-2008 22:36:48 [MindModeling@Beta] Starting task PRP_Test4-2-0.26_1.81_1.91_0.022_1202237562_1 using ACTR version 261
12-Feb-2008 22:37:01 [MindModeling@Beta] Computation for task PRP_Test4-2-0.26_1.81_1.91_0.022_1202237562_1 finished

12-Feb-2008 22:37:01 [ABC@home] Resuming task abc_wu_3996258909000_2805000_1 using abc-finder version 103
12-Feb-2008 22:37:02 [MindModeling@Beta] Started upload of PRP_Test4-2-0.26_1.81_1.91_0.022_1202237562_1_0
12-Feb-2008 22:37:04 [MindModeling@Beta] Finished upload of PRP_Test4-2-0.26_1.81_1.91_0.022_1202237562_1_0
12-Feb-2008 22:38:04 [MindModeling@Beta] Sending scheduler request: Requested by user. Requesting 0 seconds of work, reporting 1 completed tasks
12-Feb-2008 22:38:10 [MindModeling@Beta] Scheduler request succeeded: got 0 new tasks

17) Message boards : Number crunching : Trash talk (Message 361)
Posted 13 Feb 2008 by Dagorath
I think we need a sub-category for talking trash, being that in just over 24hours of work I have moved into 18th place.

w00t

Congratulations!

How does one find one\'s ranking for this project? I could not find it at the Boinc Stats site. Thanks!


Home Page in Community section, click Statistics -> Top Participants
18) Message boards : Number crunching : RESOLVED - Upload problems (Message 332)
Posted 9 Feb 2008 by Dagorath
Still getting message

09/02/2008 18:17:45|MindModeling@Beta|Message from server: Server can\'t open log file (../log_MindModeling/cgi.log)




I got the same message about 30 minutes ago when trying to attach a new host. The machine attached but when it sent the first scheduler request (to get project init files) I got:

Message from server: Server can\'t open log file (../log_MindModeling/cgi.log)

It made several more attempts over the course of 20 minutes but then I gave up and put MM on Suspend (no sense hammering server until it\'s fixed).

edit: The above was on Win XP, I\'m also getting the same on a Linux host trying to upload a result.
19) Questions and Answers : Unix/Linux : RESOLVED - 2.65/2.66 -- Or just don't blink (Message 299)
Posted 4 Feb 2008 by Dagorath
Hmmmm...

Interesting.

After a detach/reattach cycle, units are running for a measurable amount of time.

Good

I think there was a left over file in the projects directory that was created 2.64


Yes, must have been a left over or something. A simple reset didn\'t work for me but a detach/reattach fixed it. Now units are crunching error free and verifying against results from wingmen :)

Keep up the good work, Jack.

20) Questions and Answers : Unix/Linux : RESOLVED - 2.65/2.66 -- Or just don't blink (Message 257)
Posted 1 Feb 2008 by Dagorath
I can\'t really tell if they\'re working or not because they run so fast.


They seem to be working just fine, at least your log shows success for those.

I\'m getting same error as I got on 2.64, \"output file size exceeds limit\", output file is 137 MB but the limit is 50 MB.

Yours run .2 secs, mine run 2 to 3 minutes on a slightly faster machine.

Yours is 32 bit OS, mine is 64.

You have 2 GB RAM, I have < .9 GB

Top shows physical mem and swap are OK.

typical stderr out:

<core_client_version>5.10.28</core_client_version>
<![CDATA[
<message>
Maximum disk usage exceeded
</message>
<stderr_txt>
ACTR: boinc_init_options complete
ACTR: boinc_get_init_data(actr_aid) complete
ACTR: Trace 1
ACTR: Trace 2
ACTR: Trace 3
ACTR: Trace 4
ACTR: Trace 5
ACTR: Trace 6
ACTR: Trace 7
ACTR: Trace 8
ACTR: Trace 9
ACTR: Trace 10
ACTR: Trace 11

</stderr_txt>
]]>

Conclusion: takes longer to go to hell than heaven?



Next 20

Main page · Your account · Message boards


Copyright © 2021 MindModeling.org