Hi,

the server page is saying it has 159,333 Famous workunits to download and yet my manager is telling me to change my preferances as there is no work for the application i have chosen

its also sitting telling me i've reached my daily quota of 4 work units which when you consider the chip can crunch on 8 models at the same time is actually rather stupid
surely with the host of chips taking 6, 8 or even 12 wu now that figure of daily work units needs to be looked at ?

but my primary question is about the famous models or the lack there of

best regards
Ian

---

----> Please Join team Scotland HERE

Did you change your preferences as indicated? (159,316 in the queue now, so some participants are receiving work.)

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

It looks like you have 5 FAMOUS tasks running and a couple hadsm3's. The task list webpage for that computer also shows an error, and several aborted tasks in the last two days. I'm not sure what is going on. Do you have tasks suspended in the task list. More recent versions of BOINC seem to be rather temperamental on choosing when to download new tasks.

I'm using the latest 6.10.56 boic manager so i can manage each machine seperately
i do have the preferances set to only Famous models i'm only running 7 models on the i7 machine as it wont download any more and another famous model is due to end shortlyleaving me crunching only 6 models on a pc which used to be quite happy crunching 8
the new boinc manager is set to 10 days extra work and at one point had an extra 6 wu waiting to be started plus the 8 it was doing

i cant end up crunching 1 work unit because it wont download any others
i crunch on milkyway gpu project and it quite happily has 48 work units at any one time waiting to be worked on so i doubt its the manager

any help would be great
Ian

UPDATE : I just changed the prefs on the server to accept hadsm3 work units as well as famous work units and it download 4 new hadsm3 work units straight away started crunching on one waiting on the other 3
it hit the quota message or else i think it would have downloaded a few more

so its the famous models themselves it doesnt like which is really strange maybe more people will have this problem in future but it just hasnt shown up yet as two of my original 8 famous models went blue world on me also which never ever happens

best regards
Ian

P.S. Something strange has happened to the number of posts that this forum says youve made as i've been a member since 2006 and Scotland team Founder since then and have been in here posting way more than that !
my question is has the forum database lost thousands of posts or is this just a numerical error by the new forum software ?
i notice my signature doesnt show up now either !

best regards
Ian

i crunch on milkyway gpu project

GPU work uses part of a cpu to move data to and from the gpu.

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

Also, there's a new option in the latest versions of BOINC, for people who think/claim that BOINC is slowing down their other work.

Suspend work if CPU usage is above
0 means no restriction

The default setting is 25%. Setting it to zero results in the option displaying ---

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

P.S. Something strange has happened to the number of posts that this forum says youve made as i've been a member since 2006 and Scotland team Founder since then and have been in here posting way more than that !
my question is has the forum database lost thousands of posts or is this just a numerical error by the new forum software ?
i notice my signature doesnt show up now either !

Best posted in the thread intended for problems. Milo monitors that thread, and will look into problems.

Here >>> Server upgrade issues

Sometimes if you can't fetch work you should suspend work from other projects. I know this manual intervention shouldn't be necessary, but...

It means that if you have several projects you can store up work way beyond the 10-day work buffer.

I wonder whether posts before a certain date have been deleted. You certainly haven't accidentally acquired a new identity because all your credits are there. Just as well the software hasn't deleted all your credits before a certain date!

---

Cpdn news

Maybe the same problem over here. My box shows this:

14.06.2010 19:20:59 climateprediction.net Sending scheduler request: To fetch work.
14.06.2010 19:20:59 climateprediction.net Requesting new tasks for GPU
14.06.2010 19:21:04 climateprediction.net Scheduler request completed: got 0 new tasks
14.06.2010 19:21:04 climateprediction.net Message from server: No work sent
14.06.2010 19:21:04 climateprediction.net Message from server: No work available for the applications you have selected. Please check your preferences on the web site.
14.06.2010 19:21:16 climateprediction.net update requested by user
14.06.2010 19:21:20 climateprediction.net Sending scheduler request: Requested by user.
14.06.2010 19:21:20 climateprediction.net Requesting new tasks for GPU
14.06.2010 19:21:25 climateprediction.net Scheduler request completed: got 0 new tasks
14.06.2010 19:21:25 climateprediction.net Message from server: No work sent
14.06.2010 19:21:25 climateprediction.net Message from server: No work available for the applications you have selected. Please check your preferences on the web site.
14.06.2010 19:21:45 Collatz Conjecture suspended by user
14.06.2010 19:22:01 climateprediction.net update requested by user
14.06.2010 19:22:05 climateprediction.net Sending scheduler request: Requested by user.
14.06.2010 19:22:05 climateprediction.net Requesting new tasks for GPU
14.06.2010 19:22:10 climateprediction.net Scheduler request completed: got 0 new tasks
14.06.2010 19:22:10 climateprediction.net Message from server: No work sent
14.06.2010 19:22:10 climateprediction.net Message from server: No work available for the applications you have selected. Please check your preferences on the web site.
14.06.2010 19:23:15 climateprediction.net Sending scheduler request: To fetch work.
14.06.2010 19:23:15 climateprediction.net Requesting new tasks for GPU
14.06.2010 19:23:21 climateprediction.net Scheduler request completed: got 0 new tasks
14.06.2010 19:23:21 climateprediction.net Message from server: No work sent
14.06.2010 19:23:21 climateprediction.net Message from server: No work available for the applications you have selected. Please check your preferences on the web site.
14.06.2010 19:23:52 climateprediction.net update requested by user
14.06.2010 19:23:56 climateprediction.net Sending scheduler request: Requested by user.
14.06.2010 19:23:56 climateprediction.net Requesting new tasks for GPU
14.06.2010 19:24:59 climateprediction.net update requested by user
14.06.2010 19:25:01 climateprediction.net Scheduler request failed: HTTP gateway timeout
14.06.2010 19:26:01 climateprediction.net Sending scheduler request: To fetch work.
14.06.2010 19:26:01 climateprediction.net Requesting new tasks for GPU
14.06.2010 19:26:06 climateprediction.net Scheduler request completed: got 0 new tasks
14.06.2010 19:26:06 climateprediction.net Message from server: No work sent
14.06.2010 19:27:12 climateprediction.net Sending scheduler request: To fetch work.
14.06.2010 19:27:12 climateprediction.net Requesting new tasks for GPU
14.06.2010 19:27:17 climateprediction.net Scheduler request completed: got 0 new tasks
14.06.2010 19:27:17 climateprediction.net Message from server: No work sent

Workunits for GPU would be fine in the future. My boinc manager (6.10.18) joins future demands:-)

Greetings bonsai911

---

Now I'm receiving new workunits! Some strange behaviour in preferences, (new entry or options?) blocked the transfer.

Bonsai911

---

Bonsai, I see that you are from Germany. If you use the German translation, did you need to edit the ClimatePrediction preferences in your account in order to get new work?

The second item of this project section of the preferences was wrong in German when Milo upgraded CPDN's Boinc version at the beginning of June. Now it's correct. It said GPU. Now it correctly says CPU.

---

Cpdn news

Yep, that's the point. I changed that option and the workunits arrived.


Thanks

Bonsai911

---

I can't get famous units to download. Two had download errors on 21 Jun 2010 18:50:44 UTC and 14 Jun 2010 13:56:37 UTC, now only get:

6/22/2010 9:15:10 PM climateprediction.net Requesting new tasks
6/22/2010 9:15:15 PM climateprediction.net Scheduler request completed: got 0 new tasks
6/22/2010 9:15:15 PM climateprediction.net Message from server: No work sent
6/22/2010 9:15:15 PM climateprediction.net Message from server: No work available for the applications you have selected. Please check your preferences on the web site.
6/22/2010 9:15:15 PM climateprediction.net Message from server: (reached daily quota of 1 tasks).

According to the site I have a daily quota of 4 per CPU, that would be 8 total. That message also causes BOINC to defer communications for 22 hours.

Tried changing to HADAM3P but the server will not accept changes to the preferences.

Any idea when this download problem will be fixed?


EDIT
I was able to get a HADAM3P unit, will try for another famous in a few days.

---

Keep on crunching Pizza@Home

I suspect that you have 3 separate problems.
Some of it is a side effect of the server code upgrade. It's being discussed.
Some of it was due to a problem with mirroring, which has been fixed.

Tried changing to HADAM3P but the server will not accept changes to the preferences.

Not sure about this. I'll pass it on.

---

Backups: Here

Ian, I've just changed my CPDN model type preferences twice and it worked both times. First I changed from FAMOUS to HadAM3P and then back again. I presume you updated CPDN from the Boinc manager after editing your account?

However, after updating CPDN and my changed preference apparently taking effect I see that on my account page (I'm sure you'll see your own, not mine, but you'll see which page I mean) the red text at the top hasn't disappeared. It still says:

Your preferences have been updated. Client-related preferences will take effect when your computer communicates with climateprediction.net or you issue the Update command from the BOINC client.

I've updated CPDN again from the BM but the red text's still there even though I've refreshed the page, as if I hadn't yet done the project update. Surely this red text should disappear as soon as the server has activated the preferences edit?

This could indicate that a) I don't understand or remember how the red reminder's supposed to work, b) that my edit didn't really take effect, or c) that there's a problem with the red reminder in the new Boinc version that Milo upgraded to.

I can't try fetching a new model to see whether I get a FAMOUS (my most recent preference) as my computer already has far more work for CPDN than the 10-day buffer.

---

Cpdn news

1) The red text on the top of the preferences page is triggered by the extra parameter '&updated=1' on the end of the url. No amount of updating or refreshing the page will get rid of it, but if you navigate to somewhere else and reload the page from the "Taking part in CPDN" navigation pane, you'll get a copy without that parameter and without the red text.

2) Some of the settings on that page (such as Resource Share) affect the way the copy of BOINC on your computer behaves. You needs to click 'Update' in BOINC Manager to transfer those settings to your computer before they'll work.

Other settings, like the preference for which type of model you'd like to receive next, only affect the way the server behaves. You don't need to 'update' for those: the server will use the new values anyway next time they're needed.

Thank you for the insight into the usually inscrutable inner workings of Boinc! Who would have guessed your 1) comment?

If I understood the Boinc code as well as Richard I'd change my username to Sibyl at Delphi.

---

Cpdn news

If I understood the Boinc code as well as Richard I'd change my username to Sibyl at Delphi.

Weren't you in Fawlty Towers then? :P

So what is the maximum amount of "tasks" per CPU/day then here? 3? As my brand new system, only in operation since this weekend has zero. ;-)

I can explain the aborting of the one model, I can't explain the detachment.
The aborting was done as I thought that the one running was the one showing client detached. Sorry, I am not used to new "tasks" showing up in the middle or at the bottom of any page, I usually expect them at the top of the page.

---

Jord.

The default daily quota of models per CPU is 4. So that should allow the very nice computer 16 models per day in theory. But as 16 could never complete within the 10-day buffer time you could AFAIK never download so many.

Your aborted model will have reduced the quota to 3 per day ie 3 x 4 cores = 12.

I don't think anyone here truly understands all the reasons for the Client detached classification. It certainly doesn't necessarily mean that the computer was detached. It can mean you restored the model from a backup, but you probably didn't do that. On CPDN the most common description of Client detached is 'mysterious'. But it doesn't prevent a model from crunching.

I can't understand either why Client detached is called an 'Outcome' on the task page which to me means what happens at the end ie a result. But on the computer's All tasks page the same thing is called Status, which I think is the right word.

---

Cpdn news

Restoring a single task from backup shouldn't make any difference. But restoring the whole <project> section of client_state.xml, or indeed the whole file if you're only connected to one project, will mess up the RPC sequence numbers. Remember that for CPDN, each trickle will increment the RPC sequence, so merely saying "I didn't download a new task, or report a completed one" doesn't exempt you from this gotcha.

And I suspect, though don't have any evidence, that a premature sequence number is regarded by BOINC as suspicious of cheating, and is one potential cause of the 'detached' message.

I can't see any reason at all for the daily quota to have been reduced to 0. The daily default per core is still 4 per day. That's what my Intel has. My AMD only has 3 because it crashed a model in April.

[Edit]I've downloaded a new model to the AMD and its daily quota has remained at 3 ie the quota has behaved correctly. It shouldn't increase to 4 again until I upload a completed model.

The AMD's graphics card isn't usable for crunching. I have an uneasy feeling that the daily quota's going wrong with CPDN's new Boinc server version only for computers with a GPU that's usable on some other projects. This is just a feeling.

---

Cpdn news

The AMD's graphics card isn't usable for crunching. I have an uneasy feeling that the daily quota's going wrong with CPDN's new Boinc server version only for computers with a GPU that's usable on some other projects. This is just a feeling.

I got two separate scheduler requests at this moment. One for the CPU, telling me in answer there was no new work, yet 5 seconds later it downloaded two new models. The other was for the GPU, which reported me the following:

24/06/2010 17:35:32 climateprediction.net Message from server: No work sent
24/06/2010 17:35:32 climateprediction.net Message from server: No work available for the applications you have selected. Please check your preferences on the web site.
24/06/2010 17:35:32 climateprediction.net Message from server: (reached daily quota of 4 tasks)

So I'll think that's the "bad" message saying there's no work for the GPU around here. It won't keep me awake. :)

The daily quota for this machine for CPU is still zero though.

The detach... OK, let's see.
I got all parts for this machine on Tuesday, built it, found that the OS they'd given me to play with the system was Vista 32, so downloaded an official ISO for Windows 7 Pro, burned that and installed it. That was Wednesday and Thursday. Must've installed BOINC on it by Thursday, as that's where the first superfast returns sit at at Einstein (downloaded the 17th).

Saturday I got Windows 7 Ultimate, installed that and found it needed a clean installation, as the language of 7 Pro and Ultimate differed. The BOINC Data directory was at a different partition, so I clean installed Windows, reinstalled all the software including BOINC. And then I found that upon starting BOINC it would start two boinc.exe files and never actually start BOINC. The only way to fix that for me was to remove the BOINC data directory completely and reinstall BOINC clean, start all over.

I never detached as I couldn't get BOINC to start on the "old" tasks again. Yet all, even at Einstein, show client detached. So I can only figure that's it.

Sorry to "pollute" the FAMOUS thread with this, as it's hadmap3 work I am talking about. Someone move it to somewhere more convenient if necessary. :)
(hadamp3's in 3 days...? wow... /me strokes his new system)

---

Jord.