climateprediction.net home page
Posts by Robi

Posts by Robi

1) Message boards : Number crunching : Too many everything... (Message 35853)
Posted 6 Jan 2009 by Robi
Post:
There are now some other projects with fairly long tasks that would also benefit. However, as far as I know nobody has yet thought of a way to make BOINC completely exit from itself (ie stop running completely) but still remain active to back itself up. It would have to be done by BOINC itself.


I think there are some ways for XP and Vista running as BOINC service:
[code]net stop BOINC
backup
net start BOINC[/code]

or
[code]sc stop BOINC
backup
sc start BOINC[/code]

should do the trick...
2) Message boards : Number crunching : Too many everything... (Message 35850)
Posted 6 Jan 2009 by Robi
Post:
There are now some other projects with fairly long tasks that would also benefit. However, as far as I know nobody has yet thought of a way to make BOINC completely exit from itself (ie stop running completely) but still remain active to back itself up. It would have to be done by BOINC itself.


what I meant to say is, that CPDN should back up its own task(s) and upon error (if it\'s not computation but crash) restore to its last \"checkpoint\"?
Id est, let each project backup its own tasks.

yes, it would be nice if BOINC could implement an automatic backup strategy (especially for the service installation), but with my meager 3 projects, CPDN seems to be the only one with crashing problems (note: so far I haven\'t had a crash on CPDN with this system [knock on wood] although 4 systems have this far quit on me :(... )
3) Message boards : Number crunching : Too many everything... (Message 35840)
Posted 6 Jan 2009 by Robi
Post:
Before Christmas we were discussing with Milo, one of the CPDN programmers, whether it\'s possible to get rid of at least some of these phrases that mislead lots of our members. Now that the holidays are over we\'ll need to restart this discussion. The problem is that this stuff is built into the BOINC-provided web pages because most projects need it all.

To maximise your chances of completing the model, have a look at the CPDN README collections. There\'s a link in my signature. It\'s very useful to select a backup method from the collection so that if your model does crash you can restore and continue it.


Hi mo.v,
if and when you restart those discussions, it would be grand to implement a CPDN own backup strategy and not rely on users who might or rather not do backups of

  • their systems
  • BOINC
  • CPDN


or the ones that do a backup don\'t stop CPDN... and then have a crash of the model...

4) Message boards : Number crunching : Too many everything... (Message 35824)
Posted 5 Jan 2009 by Robi
Post:
That part of the software is used by other projects, but here the info is fairly meaningless. Just ignore it. The only thing of use on that page is the list of other models running, if YOU start to have a problem with yours and you want to see how others are doing.

This matter has been mentioned a lot of times, and you may be able to find these posts by using the search option at the top of this page if you want more information.


Hi Les, thank you for the explanation.

BTW I did try the search before I posted :) I also looked through the FAQs, but couldn\'t find anything that would explain it, at least I didn\'t see anything on the \"too many\" search that I entered (right now only my post shows up if I search for \"too many\") ;)

...and as I mentioned, I seem to have the ONLY model running %-/
But again, thanks for the explanation. Now I know that all I need to do is just keep it up & running :)
5) Message boards : Number crunching : Too many everything... (Message 35822)
Posted 5 Jan 2009 by Robi
Post:
I don\'t understand this WU.
It says \"errors: Too many error results Too many total results\"
On the list I seem to be the only one out of 8 that is actually doing some work on it. Everybody else had either \"client error\", 1 \"client detached\" and 2 \"didn\'t need\"?
What does that mean: \"Too many error results\"?
What does: \"Too many total results\" mean?

I suppose my work in progress will still be useful, as it\'s still chugging silently along...
6) Questions and Answers : Windows : Visual Fortran run-time error (Message 2328)
Posted 31 Aug 2004 by Robi
Post:
> HP Pavilion a562n P4 3.0E 512mb
>
> I got the same message, HP loads a bunch of garbage software on its computers,
> Delete HP's software in addremove programs or remove them from memory in task
> manager.
>
> That fixed my problem anyhow.

oh, I had several of them removed when I got the PC.

on this PC, I run Apache httpd server. (if I turn it off, you won't be able to see the error message :-/ )

--
Robi
7) Questions and Answers : Windows : Visual Fortran run-time error (Message 2252)
Posted 31 Aug 2004 by Robi
Post:
I have been running BOINC as beta test since december '02. Today I saw a message for running CPDN on BOINC and decided to run it.
now at about every percent of progress (or so), the following popup appears, and upon hitting the OK button work in progress freezes.
<img src="http://216.198.119.31/BOINC/CPDN/cpdnerr01.gif">

My PC is a HP Pavilion, 933 MHz, 256MB RAM, 30 GB free HD space.
sorry, forgot the 2nd most important part: WinME.


stderr_um.txt containd the following entries:
CLOSE: WARNING: Unit 60 Not Opened
CLOSE: WARNING: Unit 62 Not Opened
CLOSE: WARNING: Unit 63 Not Opened
CLOSE: WARNING: Unit 64 Not Opened
CLOSE: WARNING: Unit 65 Not Opened
CLOSE: WARNING: Unit 66 Not Opened
CLOSE: WARNING: Unit 67 Not Opened
forrtl: This function is not supported on this system.


<b>Update:</b>

CP started at 05:43
at about 10:00 I noticed the error. CP had about 2 hours of CPU time.
When LHC and CP "switched" at 10:43, CPU time was increasing for both projects.
Somehow CP doesn't stop when it should...

While I was writing this, LHC was "preempted", SAH still didn't receive work and CP was the only one running (and increasing CPU time).

CP had currently a CPU time of 04:37:00 at 12:20 CDT. (that's PM - with 0.02% done :)


I then clicked on "show graphics" and had the following information on the screen:
Phase : 1 of 3 / Timestep : 144 of 259248
Model Date : 04/12/1810 00:00
Run ID: 2s4w_000151086, CPU Time: 0000:16:06 (6.71 s/TS)

When I closed the error message (Visual Fortran run-time error)
the CPU time in BOINC for CP stopped.
--
Robi




©2024 climateprediction.net