I have 3 pending tasks that i mistakenly trashed.
(reset the project without aborting them first)
They\'re just sitting in limbo in my pending credit.
Should anything be done with them ????

7996973 6220912 847241 45,100.80
7996982 6220913 847241 28,304.64
7997030 6220919 847241 18,973.44

---

They still show as in progress on your tasks web pages but they\'ll later be classed as aborted, I expect. They will have disappeared from the Tasks tab of your Boinc manager. You don\'t need to do anything else. You\'ll keep the credits you earned for all the crunching you did with them.

That\'s bad luck because those three models were all progressing well. Why did you want to abort them?

Have you by chance got a fairly recent backup of the contents of your BOINC folder? If you have, you could restore it and get these three models back.

---

Cpdn news

Yeah, I stupidly hit the reset project button when those newer
CM3\'s started coming in and crashing. I really like running
these longer models to the end ..... Ed

---

Hi,

I had to reinstall my system, with a backup on an external hard drive. When I reinstall Boinc, I start joining the project again and it had new tasks. Problem is I still had one task running, meant to finish in Feb 09 (task ID 7252322) and I would like to carry on with it. Any idea how I can get it back in Boinc, knowing I have all the data on another hard drive.
Any idea on that ?

Bonjour Le Pommier

I think your previous model was running with BOINC v.6 and your computer again has BOINC v.6.

First complete your new model, but set all your projects to No New Tasks. When your new model has finished and reported and you have no tasks in your BOINC manager, exit from BOINC and delete the complete contents of your BOINC Data folder. Then restore the backup of the contents of your old Data folder into the empty new Data folder. Start BOINC again and I think your previous model should run.

If necessary you can look at the detailed instructions about how to restore a backup in the CPDN READMEs (link in my signature).

Don\'t worry about the model\'s completion deadline; the CPDN servers ignore deadlines. And yes, it would be a good idea to complete this model which had almost finished phase 2.

---

Cpdn news

My Climate Prediction 200 hours of work are lost after stopping the BOINC and restarting the Mac Mini. The mini was working 24/7 and I needed to restart it for some reasons. Later I tried to run the Climate Prediction for just couple of days, then Suspend Activity of the project, do a hour pause and restart the mini and worked hour were lost again. Remaining (estimated) hours returns to start value (888 hours) every time I do restart my system and start the project again. The project can't be saved to the HDD.

The mini must be shut down soon because of planned electricity repair works and I afraid my WUs will be lost again.

Please help urgently!

---------------------------------------------------------------------------------------------------------------------
My specs are:
BOINC version when 200 hours of work are lost - BOINC 6.12.xxxxx (don't remember)
BOINC version currently - BOINC 7.0.31 vx Widgets version 2.8.10
Mac OS X 10.4.11
Mac Mini Intel CoreDuo 1.66
1536 MB of RAM
Remaining (estimated) hours currently - 718 hours

The Coupled Ocean models don't like being interrupted at certain times.
But I think that in your case, it's not being allowed to run for long enough to reach a checkpoint, and is returning each time to the only place it knows about, the beginning.

200 hours should be long enough to reach several checkpoints. However, BOINC only runs it's work when you aren't using the computer, and you if use it heavily then the projects won't get much time for their work. And, as you appear to be running several projects, the work from each will have to share what time is being allowed.

This project won't suit everyone, and perhaps that includes you, because of your computer usage.

---

Backups: Here

I think these Hadcm coupled models checkpoint after each 6 model days.

As you say the models go back to the begining every time you restart the machine, I wonder whether there's something wrong with the way you're closing down the system. I know you said you stop BOINC before shutting down the machine, but are you just closing the BOINC Manager window? Or are you exiting from the whole BOINC process?

I first suspend the models using the Activity menu. You can then exit from BOINC in either of two ways:

In BOINC Manager > File > Exit

Right-click on the BOINC icon > in the menu select Exit

The tasks from a lot of projects don't seem to mind the computer being shut down while BOINC is still processing but climate models don't like this at all.

---

Cpdn news

Here is the Mac Mini and its Hadcm is here.

These models send a trickle to the server after each model year and each trickle is recorded on the task page. Your model hasn't sent a single trickle despite crunching for 200 hours...

I see that the Mini has a 1.66GHz CPU and two cores.

The first computer I crunched one of these Hadcm models on (it was a BBC model) had a 1.33GHz CPU. The minimum spec that CPDN required for these models was 1.6GHz so my computer was below the minimum spec. It could still crunch the model but it was really slow.

BUT my old slow desktop only had one core. Yours has two. Are you crunching tasks from another BOINC project simultaneously on the other core? I think this would be too much for the machine and if this is what's happening it may explain why the model's never reaching the end of its first model year.

It looks to me as if the Mini has enough RAM and this wouldn't be a limiting factor unless the other core is crunching very memory-intensive tasks.

If the scenario I'm imagining is true, I think you'd have to set all your projects to No New Tasks in the Projects tab and see whether this model can then progress as it should.

I don't know very much about Mac Minis but I also wonder whether their cooling system is powerful enough to crunch BOINC on two or even one core if the CPU is being used at 100%. I wonder whether the machine is throttling ie limiting CPU usage to keep the CPU cool. If this is the case would it be necessary to limit CPU usage in account preferences?

---

Cpdn news

However, BOINC only runs it's work when you aren't using the computer, and you if use it heavily then the projects won't get much time for their work. And, as you appear to be running several projects, the work from each will have to share what time is being allowed.

I have set somethig like Run BOINC Always in my Prefs, and it runs always as my CPU activity utility shows. I tried to run the only Climate Prediction project under the BOINC to leave the CPU absolutely free for it as well. As result, two weekends work were wasted after monday morning system restart. Seems like I should check the same with BOINC 7.0.31, maybe it has no such a feature.

...but are you just closing the BOINC Manager window? Or are you exiting from the whole BOINC process?

I first suspend the models using the Activity menu. You can then exit from BOINC in either of two ways:

In BOINC Manager > File > Exit

I was afraid to lose several hours work, so I did following:

1 Activity > Suspend (don't remember if I've seen that time the CPU load reached zero level. Mostly - yes)
2 File > Exit
3 Apple Menu > Restart

The tasks from a lot of projects don't seem to mind the computer being shut down while BOINC is still processing but climate models don't like this at all.

I tried Activity-Suspend in the BOINC menu, but both CPU cores of my mini were still loaded for few hours though.

...Are you crunching tasks from another BOINC project simultaneously on the other core? I think this would be too much for the machine and if this is what's happening it may explain why the model's never reaching the end of its first model year.

I ran only the Climate Prediction prj and the CPU was loaded for about 55% (55% for each core)

If the scenario I'm imagining is true, I think you'd have to set all your projects to No New Tasks in the Projects tab and see whether this model can then progress as it should.

I'll try this.

I don't know very much about Mac Minis but I also wonder whether their cooling system is powerful enough to crunch BOINC on two or even one core if the CPU is being used at 100%. I wonder whether the machine is throttling ie limiting CPU usage to keep the CPU cool. If this is the case would it be necessary to limit CPU usage in account preferences?

The mini is located on the west-directed window-sill and huge window frame cracks with -15 Celsius outside the window make the CPU in just perfect condition :-) Also, the CPU Activity Monitor shows no throttling signs if I even crunch simultaneously 2 projects with 100$ CPU load.

So what would be the safe way to shut down my Mac mini tomorrow?
How can I check if the Climate Prediction has saved at least one checkpoint so I could suspend it and quit the BOINC client before shutting down?

I have set somethig like Run BOINC Always in my Prefs, and it runs always as my CPU activity utility shows

That "always" is only while YOU aren't typing or running something else. i.e. BOINC only runs in the gaps between pressing a key.

How can I check if the Climate Prediction has saved at least one checkpoint

There's a countdown timer which is visible in the Show graphics display. When it reaches zero, the disk activity light should flash for a while as data is saved.

(In the Tasks tab, click on the model name, then click the Show graphics button.)

---

Backups: Here

I have set somethig like Run BOINC Always in my Prefs, and it runs always as my CPU activity utility shows

That "always" is only while YOU aren't typing or running something else. i.e. BOINC only runs in the gaps between pressing a key.

If I type many letters or move my mouse for much time the CPU Activity Monitor bar does not fall to zero level like it was few years ago: the CPU started to crunch in a minute of user inactivity. Now, if I even suspend, CPU left loaded for very long time-hour or more. My PC BOINC reacts in just few seconds when got suspended or resumed!

There's a countdown timer which is visible in the Show graphics display. When it reaches zero, the disk activity light should flash for a while as data is saved.

(In the Tasks tab, click on the model name, then click the Show graphics button.)

Neither BOINC 6.12 version nor current 7.0.31 do not show graphics of any project. The MAc minis have no HDD LEDs.

Neither BOINC 6.12 version nor current 7.0.31 do not show graphics of any project. The MAc minis have no HDD LEDs.

No graphics window appearing after clicking the Show Graphics tab of any project under Mac OS 10.4.11. Can I catch the checkpoint with other way?

What a nuisance. I've looked at what I think is the same Mac Mini on Rosetta and SETI and noticed some interesting things. But I can't post what I wrote because the Akismet anti-spam program for this forum is blocking my post. Of course there is no spam in my post.

I'm going to ask Jonathan to take Akismet off in the morning.

---

Cpdn news

I think that there's something in cc_config.xml to save data; one of the debug flags, perhaps.
I've never tried that, because Show graphics has worked perfectly for years.

If you select (click on), the model name, does the show graphics button become active, or does it remain greyed out?
If it's active, then there's a problem with one or more of the files on your computer needed to display the OpenCL graphics.

---

Backups: Here

I've just been looking at one of my hadcm3 models, and it's check pointing about every 2 minutes.
It's also reporting about 500 hours to complete.

Your model may well be encountering a different problem, but the information needed is in the Show graphics window.

---

Backups: Here