The fact that the thread title includes 'FAMOUS' doesn't matter much. The real problem for some computers is work fetch. Problems with this could probably affect any model type. Scottish webcams may have thought the problem was specific to FAMOUS, but I don't think anyone's identified it as model-specific at all.

---

Cpdn news

The fact that the thread title includes 'FAMOUS' doesn't matter much. The real problem for some computers is work fetch. Problems with this could probably affect any model type. Scottish webcams may have thought the problem was specific to FAMOUS, but I don't think anyone's identified it as model-specific at all.

I have no problem with receiving HADAM3P models, but can not get a FAMOUS model so at least to me it appears to be model-specific. My AMD has a daily quota of 4 and even after getting the message that I reached my quota of one for FAMOUS I can get a HADAM3P with no trouble.

Here is my last attempt to get one while reporting a completed HADAM3P

6/24/2010 4:02:29 PM climateprediction.net Sending scheduler request: To fetch work.
6/24/2010 4:02:29 PM climateprediction.net Reporting 1 completed tasks, requesting new tasks
6/24/2010 4:02:34 PM climateprediction.net Scheduler request completed: got 0 new tasks
6/24/2010 4:02:34 PM climateprediction.net Message from server: No work sent
6/24/2010 4:02:34 PM climateprediction.net Message from server: No work available for the applications you have selected. Please check your preferences on the web site.
6/24/2010 4:02:34 PM climateprediction.net Message from server: (reached daily quota of 1 tasks)

I do have another HADAM3P that I am running.

---

Keep on crunching Pizza@Home

Thanks Ray for both the comments and the messages extract.

The other day Thyme Lawn explained something to the other moderators that I didn't know before.

When you look at the Server status page you see thousands of models available. Apparently plenty of each type for whoever selects them.

However, the Boinc server program that gives out the models maintains a queue of 100 available models/tasks. The rest of all those models aren't in the available queue; they're waiting to enter it. It could be that at any one time most or all of the models in the queue are of one type, or of the two types you haven't selected.

So that message really means that there isn't a model of your selected type in the current queue.

So what you say doesn't (yet!) convince me that it's a FAMOUS-specific problem. It could be because of the way the queue works. In some cases there may also be a problem with how CPDN's new server version is dealing with some computers' daily quota.

In the case of computers with SSE2 or SSE3, members will increase their chances of receiving work immediately if they select both HadAM3P and FAMOUS should they prefer a shortish model. That leaves the HadSM for members who prefer longer models or whose computers only have SSE and can only crunch this model type.

Otherwise you may just have to wait for a while hoping that the mix of model types in the queue will change.

---

Cpdn news

Thanks Ray for both the comments and the messages extract.

The other day Thyme Lawn explained something to the other moderators that I didn't know before.

When you look at the Server status page you see thousands of models available. Apparently plenty of each type for whoever selects them.

However, the Boinc server program that gives out the models maintains a queue of 100 available models/tasks. The rest of all those models aren't in the available queue; they're waiting to enter it. It could be that at any one time most or all of the models in the queue are of one type, or of the two types you haven't selected.

So that message really means that there isn't a model of your selected type in the current queue.

The Feeder being empty could have been an option, but "Message from server: (reached daily quota of 1 tasks)" normally means any other error-messages is wrong.

Since computer has reached the daily quota, even if there is work available in feeder, the computer wouldn't get any.

I am still getting the same message when trying to get FAMOUS models. This started when Milo was moving the upload servers around.

Ingleside, this is a FAMOUS problem, it dues not effect the other models. There are a few of us with this, and we can get all the models we want of other types.

I updated BOINC from 6.10.18 to 6.10.56 with no change in being able to get FAMOUS models.

Houst having problems is 1053982

---

Keep on crunching Pizza@Home

If Boinc is behaving badly I think it will be the new version Milo installed on the CPDN server on 3 June. I think 6.10.56 and other Boinc versions for our home computers must be OK for work fetch.

But I've said 'I think' twice. I don't think the daily quota mechanism is working correctly for all computers. For example, I can see no reason why the quota for Ageless's new computer was reduced to 0 when he had only downloaded 3 models, one of which he aborted. And he got a new model in spite of the 0 quota without needing to wait until the next day.

---

Cpdn news

The problem discussed here in the posts of 26 June may be related.

---

Cpdn news

I am still getting the same message when trying to get FAMOUS models. This started when Milo was moving the upload servers around.

Ingleside, this is a FAMOUS problem, it dues not effect the other models. There are a few of us with this, and we can get all the models we want of other types.

I updated BOINC from 6.10.18 to 6.10.56 with no change in being able to get FAMOUS models.

Houst having problems is 1053982

I was only pointing-out that if you get "reached daily quota", any accompanying message of "no work available for FAMOUS" most likely is incorrect, and the Feeder does have some FAMOUS available...

As for why you'll get "no work available" if you don't get a "reached daily quota", my explanation in this thread should probably be good enough. ;)



But anyway, as for your problems... With the resent server-upgrades, chances are CPDN has started using the new quota-system, and while the quota before was per BOINC-project, it is now Per application-version. So, you'll have one daily quota for FAMOUS, and another daily quota for Hadam3P, and a third daily quota for HADSM3. So, even if you've reached the FAMOUS-quota, it doesn't mean you've reached the Hadam3P-quota, and can get Hadam3P-work.

Also, the new code includes various bugs,among these a feeder-bug that leads to garbage being used in database, and can be part of the reason for incorrect quota, see changeset-21812

That sounds like what could have happened, easily. Now if Milo could let the server know that I can get that one allowed it would be good. With not being able to get even that one I started FAMOUS on another of my slower systems.

Is it possable that on the 2nd one to get a download error on 21 June it was while the server was being working on, which caused mine to to flagged to not get and FAMOUS jobs? At least the computer is happy with running an AM3P.

---

Keep on crunching Pizza@Home

Milo knows about the problems being reported here.
The difficulty is with the new server code, and some of the side effects.

For instance, manually changing the quota doesn't work, as other parts of the code change it back.
Everyone is just going to have to be wait until someone figures out how to nail reality firmly back in place so that it can't keep reverting to unreality again.

---

Backups: Here

This computer is working, but I can't download any jobs because my quota is zero.

Other than the two jobs I had to abort, the rest of the errors were not my fault. I'm almost done with the current model 82% done, but still have 1 idle CPU. How can I get my quota to at least 1???

Boinc never tries to judge whose fault failed tasks are. It just knows they haven't completed and reduces the quota.

A computer's quota should as far as I know only reduce to 0 if it's already downloaded all the tasks it's allowed for that day or if it's just crashed a lot of tasks. I believe it should start every new day with a minimum quota of 1 task per core. This means the maximum time a core should be idle with no work should be until the next day your time when the quota's reset.

Your computer's quota isn't doing this. The computer hasn't downloaded any models for ages and hasn't crashed any tasks today. Yet its quota is 0.

I don't know why. The server's new Boinc version isn't behaving as anyone would expect on certain computers.

Set the computer's buffer to 10 days. In the climateprediction preferences of your account say Yes to every model type and that you'll accept work from other models if available. Yes to everything. Then update CPDN in the computer's Boinc manager.

If these ideas have no effect on the quota please let us know (you've probably tried them already).

---

Cpdn news

I have an idle CPU; therefore, it should let this CPU have at least 1 WU. Tried your settings, but I get the same result.

12-Jul-2010 12:14:31 [climateprediction.net] update requested by user
12-Jul-2010 12:14:31 [---] [work_fetch_debug] Request work fetch: project updated by user
12-Jul-2010 12:14:32 [---] [work_fetch_debug] Request work fetch: Backoff ended for climateprediction.net
12-Jul-2010 12:14:34 [climateprediction.net] chosen: CPU idle instance
12-Jul-2010 12:14:34 [---] [wfd] ------- start work fetch state -------
12-Jul-2010 12:14:34 [---] [wfd] target work buffer: 86400.00 + 0.00 sec
12-Jul-2010 12:14:34 [---] [wfd] CPU: shortfall 86400.00 nidle 1.00 saturated 0.00 busy 0.00 RS fetchable 100.00 runnable 100.00
12-Jul-2010 12:14:34 [climateprediction.net] [wfd] CPU: fetch share 1.00 debt 0.00 backoff dt 0.00 int 0.00
12-Jul-2010 12:14:34 [climateprediction.net] [wfd] overall_debt 0
12-Jul-2010 12:14:34 [---] [wfd] ------- end work fetch state -------
12-Jul-2010 12:14:34 [climateprediction.net] [wfd] request: 86400.00 sec CPU (86400.00 sec, 1)
12-Jul-2010 12:14:34 [climateprediction.net] Sending scheduler request: Requested by user.
12-Jul-2010 12:14:34 [climateprediction.net] Requesting new tasks
12-Jul-2010 12:14:39 [climateprediction.net] Scheduler request completed: got 0 new tasks
12-Jul-2010 12:14:39 [climateprediction.net] Message from server: No work sent
12-Jul-2010 12:14:39 [climateprediction.net] Message from server: (reached daily quota of 1 tasks)
12-Jul-2010 12:14:39 [climateprediction.net] [wfd] backing off CPU 3 sec
12-Jul-2010 12:14:39 [---] [work_fetch_debug] Request work fetch: RPC complete
12-Jul-2010 12:14:44 [---] [wfd] ------- start work fetch state -------
12-Jul-2010 12:14:44 [---] [wfd] target work buffer: 86400.00 + 0.00 sec
12-Jul-2010 12:14:44 [---] [wfd] CPU: shortfall 86400.00 nidle 1.00 saturated 0.00 busy 0.00 RS fetchable 0.00 runnable 100.00
12-Jul-2010 12:14:44 [climateprediction.net] [wfd] CPU: fetch share 0.00 debt 0.00 backoff dt 0.00 int 60.00 (comm deferred)
12-Jul-2010 12:14:44 [climateprediction.net] [wfd] overall_debt 0
12-Jul-2010 12:14:44 [---] [wfd] ------- end work fetch state -------
12-Jul-2010 12:14:44 [---] [wfd] No project chosen for work fetch

That's interesting. The computer's web page says its quota is 0 but one of the messages says:

12-Jul-2010 12:14:39 [climateprediction.net] Message from server: (reached daily quota of 1 tasks)

One part of Boinc doesn't know what another part's doing. Perhaps there's a conflict somewhere or information isn't being transferred within Boinc. Or something else completely because both the quota and the message are wrong.

Thanks for those debug messages.

Please look at your private messages.

---

Cpdn news

I suspect that the new server code hasn't been fully/properly customised yet.
What is shown on user's computer pages as Maximum daily WU quota per CPU probably no longer applies, and is just a left over from the previous version.

The new code has the quota set per application, and this isn't being reflected in the page text.
And this may or may not have been included in the many options from which to choose during install.

Testing of the code that runs on peoples computers is one thing; just a matter of lots of people testing it. But there's less than one hundred active projects, and not all of them have the time or spare hardware to test new server code. It's possible that this only gets tested on SETI, which has different requirements to this project, and so parts of the code may not get tested until it's installed here.

The best advice at the moment is probably: Keep trying, and good luck.

---

Backups: Here

Did you guys fix something? I got 1 new task today at 12:00 UTC. Strangely, my quota is still at 0/day.

Over at the SETI forums, they have implemented an additional quota system, but I'm unfamiliar with how it works. Perhaps this quota system is already in the CPDN codebase.

Some of my hosts started suck air:
19/07/2010 11:03:43 climateprediction.net Message from server: No work sent
19/07/2010 11:03:43 climateprediction.net Message from server: (reached daily quota of 1 tasks)
This situation is never seen before (for me). I am discouraged.

---

Dear Metalius

Since the recent upgrade this has been happening to a lot of people. In most cases the server will give you a new WU after several hours or a day or two of waiting. One thing that help is to set your preferences to keep enough work on your computer for 10 days (the maximum). That way your computer will start asking for work several days before you complete the tasks that you are running now. This is the best way to keep all your cores busy all the time.

---

Metalius, if you or other members cannot get enough CPDN models at the moment to keep your computers busy you could if you wish join the CPDN Beta project and download models from there. On Beta there is no work fetch problem.

When Milo solves the problems with the CPDN Boinc server version you could either continue to process models for both projects or return to CPDN only.

You would need to send me a private message and I would explain how to join Beta.

Please don't be discouraged. This isn't Milo's fault and the delays in the fixing of these problems are because he has far too much work for one person. Tolu also has much more work with the models than one person should have.

It is also possible to keep computers busy by fetching work from other projects eg Rosetta and Einstein. There are so many useful projects that no computer needs to have idle cores.

---

Cpdn news

Dear friends,
Thank You for replies and advices. I am glad to see - the project team has plans to fix this problem.

---