climateprediction.net home page
Lost jobs and "busted" link

Lost jobs and "busted" link

Questions and Answers : Windows : Lost jobs and "busted" link
Message board moderation

To post messages, you must log in.

AuthorMessage
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 6952
Credit: 20,843,205
RAC: 0
Message 60652 - Posted: 12 Jul 2019, 20:16:35 UTC
Last modified: 12 Jul 2019, 20:47:24 UTC

Post from James Mayor.
Re-posted to remove sensitive info.

*****************



I have been using BOINC linked to you for a while, but I seem to be having trouble. Several times I seem to lose the particular job my computer is working. What usually happens is I see that I have several days work left on an item, but the following day the job is gone. There is no way I can trace it or determine what happened. I only “subscribe” to you (no other projects) and I have two computers running 24/7, and no other unusual application running at the same time. I have noticed several jobs that have been running for many days, up to maybe eight or nine, just disappear. That is a significant waste.

Do you have any suggestions?

Jim mayor


P. S. Incidentally, I attempted to go to the forum (You can ask BOINC-specific questions on the climateprediction.net Forum. at the bottom of this page: https://www.climateprediction.net/support/technical-faq/#Questions_Problems_and_Discussion_Forums
The link takes me to http://climateapps2.oerc.ox.ac.uk/cpdnboinc/forum_index.php which is “Not Found.”
ID: 60652 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 6952
Credit: 20,843,205
RAC: 0
Message 60653 - Posted: 12 Jul 2019, 20:28:19 UTC

Hello Jim.

Please, no email addresses.

I'm not sure why so many of your models are failing, as there's no real error message.

Two possiblities:
Not enough memory
A sudden interuption of the running

These climate models don't like such things as turning off the computer while they're running.
Always use the BOINC menu options to suspend BOINC first, then wait a few seconds to allow all of the open files to be closed and saved, then Menu -> Exit, THEN turn off / hibernate the computer.

And if you start running anything else, which uses a lot of memory, even changing the display to full screen, this may cause problems.

Thanks for the broken link. I'll pass it on.
But you got here to the forum in the end. :)
ID: 60653 · Report as offensive     Reply Quote
James Mayor

Send message
Joined: 18 May 10
Posts: 5
Credit: 2,180,348
RAC: 26
Message 60654 - Posted: 12 Jul 2019, 22:04:26 UTC - in response to Message 60653.  

Sorry about the email.

So do you receive error messages?

I think I have plenty of memory since I am not running anything else.

I can't always control the computer shutting down. MS and AVG sometimes shut it down at night. Maybe I should unplug the LAN? I don't always check, but it would seem that most of the time the computer is shut down at night, when I reboot in the AM BOINC starts okay, i. e., without losing the job. When a job runs for ten or so days I will notice that.

If I shut down the LAN near the end of a long job, will BOINC just wait until the internet is available again to upload the results? If so, maybe I'll do that.

Yes, I did find you.
ID: 60654 · Report as offensive     Reply Quote
WB8ILI

Send message
Joined: 1 Sep 04
Posts: 130
Credit: 48,738,424
RAC: 234
Message 60655 - Posted: 13 Jul 2019, 1:20:21 UTC - in response to Message 60654.  

James -

I assume you are referring to your AMD (cpu) computer.

Of the tasks that you did complete successfully, they are being suspended A LOT! CPDN tasks don't like being suspended.
Go to Tasks -> Computing preferences and UNCHECK all the "suspend" options.

CPDN tasks also have a poor record (improved recently) of restarting successfully after a re-boot, etc. But, if you have to shut your computer down each evening, I guess there is not much you do about that.
ID: 60655 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 6952
Credit: 20,843,205
RAC: 0
Message 60656 - Posted: 13 Jul 2019, 6:15:22 UTC - in response to Message 60654.  

Yes, one does get error message, and they are of many different types.

This is one that one of my computers got part way through.

Signal 11, or SigSev, is tricky to resolve, as it seems to have a few possible causes.

I haven't run real Windows for several years.
My computers run Linux, and a program called WINE, which allows me to run a "sort of" Windows. But it's written using only code in the public domain.
Which means that MS doesn't know about it and try to "update" it at their own whim.
And the version of Windows to be run can be chosen from almost the full range of versions, from back at about 2.0, up to 10. I use the XP version, because I ran the real version for so many years.

These climate models have a lot of files open all of the time, perhaps a dozen, or maybe more.
All of these need to be saved to the HD and closed, before BOINC shuts down, otherwise you can get problems. i.e. When it starts up again, it finds that some of the saved files are from one point in time, and the rest are from an earlier time. So the climate model fails.

Shutting down / unplugging the LAN would stop external forces from causing problems. But when to do this?
You can get interruptions at any time, even near the start.

There are settings in Windows somewhere that allow you to control when MS can do their updating, and the long established rule of "never let an AV program scan any part of BOINC", usually stops any problems from that source.

Keep trying things.
We're here for as long as it takes.
ID: 60656 · Report as offensive     Reply Quote
Profile Alan K

Send message
Joined: 22 Feb 06
Posts: 292
Credit: 14,738,424
RAC: 184
Message 60660 - Posted: 13 Jul 2019, 22:01:52 UTC - in response to Message 60656.  

There are settings in Windows somewhere that allow you to control when MS can do their updating,


If you do use older versions of Windows (pre Win10) then be aware that udpdate support for anything XP and earlier does no longer exist. This does have the advantage of not being plagued by "your computer needs updates" messages. Win7 support I am not sure of, however it did give you the option of checking if there were any updates and also installing them individually. The default for Win10 is "we will do everything for you", though there are workarounds for this but you need to be confident of making changes to the registry. Search the net for answers to this.

If you are worried about LAN activity there are settings in BOINC preferences that allow you to only use the network at certain times of the day (or even not at all).
ID: 60660 · Report as offensive     Reply Quote
James Mayor

Send message
Joined: 18 May 10
Posts: 5
Credit: 2,180,348
RAC: 26
Message 60683 - Posted: 20 Jul 2019, 11:40:28 UTC - in response to Message 60655.  

One of my 24/7 computer uses an Intel core i5-3470 cpu, and the other an AMD A6 Pro-7400B cpu. I have the same "problem" with both.

None of the "suspended" options are checked and I have never voluntarily suspended any job on BOINC. I never shut down the computers, but MS does.
ID: 60683 · Report as offensive     Reply Quote
James Mayor

Send message
Joined: 18 May 10
Posts: 5
Credit: 2,180,348
RAC: 26
Message 60684 - Posted: 20 Jul 2019, 11:44:03 UTC - in response to Message 60660.  

Everything is always up to date, security being what people fear.

I'll just continue on the way I am. I apologize to everyone for messing everything up. If I continue to cause problems just drop me.
ID: 60684 · Report as offensive     Reply Quote
Profile Iain Inglis
Volunteer moderator

Send message
Joined: 16 Jan 10
Posts: 979
Credit: 3,103,609
RAC: 95
Message 60686 - Posted: 20 Jul 2019, 13:41:40 UTC - in response to Message 60683.  

... None of the "suspended" options are checked and I have never voluntarily suspended any job on BOINC. I never shut down the computers, but MS does.


Just to make sure we're talking about the same thing. The important suspend option is the "Suspend when computer is in use" setting on the "Computing" tab of the "Computing preferences" dialog in BOINC Manager. If that option is set then the model will suspend when something else is happening on the computer (i.e. by default BOINC tries not to interfere with normal use of the PC). However, CPDN models don't always respond well to such suspensions, so if you don't mind the CPU being used pretty much 100% by CPDN then unset that option. You can always reduce the number of CPUs in a multiple-CPU PC to ensure that the PC responds well in normal domestic use - that's what I do.
ID: 60686 · Report as offensive     Reply Quote
James Mayor

Send message
Joined: 18 May 10
Posts: 5
Credit: 2,180,348
RAC: 26
Message 60705 - Posted: 23 Jul 2019, 23:57:14 UTC

Okay. I concede. I found that no new work was available on Climate Prediction so I opened several other projects. After I downloaded several jobs I opted out of those projects hoping that Climate Prediction would be available later. It seems today I downloaded jobs on my two computers. I opened BOINC on one computer just to see what was available. One computer showed the job running, but that computer had already opted out of the other projects so I left it. On the other computer I opened BOINC to see what was running and saw that a Climate Prediction job was running. When I went to the menu to opt out of the other projects, then went back to the main BOINC screen there was no job. I waited and was advised that no jobs were available. I believe there is a problem and I don’t think it is on my computers.
ID: 60705 · Report as offensive     Reply Quote

Questions and Answers : Windows : Lost jobs and "busted" link

©2019 climateprediction.net