I\'m new to Boinc and I\'m not sure if my projects are reporting any work. I\'m running Vista on a 1.7 Ghz laptop. I joined last week and I keep getting \"communicationn deferred\" in the status line of the project I\'m running. I\'m getting the messages below. Am I communicating work and if so, how do I get more work instead of once a day?
\'Requesting 60480 seconds of new work and reporting 2 completed tasks\"
\'Scheduler RPC succeeded [server version 509]\'
Message form server: No work sent\'
\'Message from server (reached daily quota of 2 results)
\'Deferring communication for 19 hr 55 mn\'
Reason: Requested by project\"
Is this OK?
Thanks

---

Sorry old chap, but no, it\'s NOT OK.

One short model will last for a month or more on your machine, and a long model for 6 months or so.
If you\'re \"requesting new work\" several times a day, then you\'re crashing the models.
The limit of 2 is to stop the waste of datasets by faulty hardware.
(If the models WERE working OK, then you would be uploading a \"trickle\" file to the server once-to-a-few-times a day. THIS would be normal.)


You can see the failures on your page of Results here.
Clicking on the Result ID of 3 at random shows this error: exit code 1073807364 (0x40010004)

This is a Windows \'Stop\' error message, and is usually associated with a display problem, such as the model trying to compete with some other program that uses a lot of hardware resouces.
There are some hints and tips here, in the 5 READMEs, starting with the links numbers 5 & 6, in README - Crashes and other problems

Post back if you need more help, or even if you just solve the problem. We like to know.

---

Backups: Here

Thanks for the reply.
So I think my problem is that I\'m running the models constantly even when the computer is in use so I\'ve changed that setting. I noticed that I have picked up some credits already so something must be working right somewhere. I\'ve been keeping an eye on the CPU usage and it appears to be working OK but after a while it does max out and I have to restart the computer to get to run normally. I am running dual processors with one processor being used for the models.
Is there a way I can find out what other programs might interfering?

---

LOTS of us run their models constantly. It gets more done that way.

Interfering:
1) Look at the date/time each model failed, and then think about what else was running at that time.
2) Look at Task Manager (control-alt-delete) to see what else is running.

The essential thing is to go to the link to the README posts that Les recommended. Those two README posts are specifically to help solve the sort of model crashes you\'ve had. I\'d also recommend reading the top tips in the README called Running the model. (Don\'t try to read all the READMEs, just the relevant parts we\'re pointing out.) There\'s another link to them in my signature.

It would also be a good idea to disable the screensaver as running it can cause graphics problems on some computers. You can still watch your globe when you want to through the button in your boinc manager.

When you\'ve got the hang of avoiding your model crashing (which means you then have to start afresh with a new model), your computer should do well with climate models as it crunches the data fast and actually has enough memory to run 2 models side-by-side.

When you\'ve had time to study those posts and you manage to keep a model going for several trickles, ie several model years, if you post back again we\'ll explain how you can back the model up.

---

Cpdn news

From what I\'m reading there seems to be two possible causes.
Either my graphics card isn\'t up to it or my scheduled tasks such as scanning and defragging could be interfering. I\'m going to change some settings on my computer to coordinate when programs are running to avoid conflicts. I\'ve changed a couple of settings already and the I\'m seeing the model run. I\'m going to wait this model out and see how it\'s reporting before before moving on to the next possible solution.
I did notice that while I was running Climateprediction alone that there were two modules running side by side. Now that I\'ve added other projects, only one running model is being run.

AV scans with the model running aren\'t a good idea, as you\'ve gathered, unless you exclude the model from the scan. There\'s a README post, I think by Marj, to help with how to exclude it.

You probably shouldn\'t need to defrag every day. For most people, once a week or fortnight is enough. There\'s a discussion going on at the boinc_dev forum about whether defrags with workunits running could crash them. I don\'t recollect any reports on the cpdn forums of such crashes.

Updating the graphics card drivers has helped some people avoid further 107 crashes.

Generally speaking, the precautions we recommend on cpdn to prevent workunit crashes also apply to WUs from other projects as the principles are the same.

---

Cpdn news

Thanks for all the help.
Unfortunately, I\'m going to have to wait a while before I can upgrade my graphics card.
It looks like everything is running normally and all the models are making progress and communicating results. In two days of running climateprediction without any crashes I\'ve completed 14 model months but haven\'t seen any trickles yet, but from what I\'m reading there seems to be some lag time in reporting.
The posts have been really helpful. Thanks again.

It\'s not the physical CARD that should be upgraded, just the version of the SOFTWARE it\'s running.
This will be available for download from the web site of the card\'s maker.

Also make sure that you have the latest version of DirectX, which is a graphics program.

---

Backups: Here

If you don\'t know what sort of graphics card (or anything else, hardware or software) you have you can download the freebie Belarc Advisor which within about 8 seconds produces a single list of what your computer consists of. The make of the graphics card will be listed under \'Display\'. I expect it will be ATI or NVidia.

http://www.belarc.com/ba5.html?B

---

Cpdn news

Les & MO.V
The video Card is ATI and I\'m running DirectX 9.0 or better and I upgraded the graphics driver and a couple of other drivers for display, and communications. Everything seems to be running much better.
I\'ve completed .15% on one module with a total of 4 3/4 hrs CPU time. I\'ve had the computer shutdown for the last 24 hrs, so no work since yesterday. I\'m going to keep the current settings until I start to see trickles and then raise up the settings like running constantly until I\'m satisfied I\'m not going to crash anymore.
Thanks again for both your help. I was getting a bit frustrated with seeing modules run and not get results. This has been a great help.
M

---

The other thing to do to make it much more likely you\'ll get to the end of your model/s is to regularly back up the contents of the boinc folder. Through my sig go to the first item in the README about avoiding model crashes. Les explains an easy backup method, plus how to restore the backup in the event that the model fails. Restoring a backup is the only way to continue crunching a crashed model.

---

Cpdn news