MM

Grizz said: CPDN has nothing written in the \"status\" section on the projects page. This means that BOINC is Allowing New tasks in the projects tab.
If it was disabled the status would say: Won\'t get new tasks

Earlier, Grizz posted: 26/01/2008 1:30:54 PM|climateprediction.net|(not requesting new work or reporting completed tasks)
This is a Messages tab message.

So, BOINC is
A) Allowing new tasks, and
B) Not requesting more work

Which seems to indicate that BOINC has decided that it has enough work already, given
A) The amount of work on hand from other projects
B) The resource share for cpdn (i.e. time it\'s allowed to devote to a climate model) and
C) The deadline for a climate model. As his copy of BOINC knows (from past experience, i.e models already run), how long a climate model will take to complete, it\'s probably decided that there\'s no point in requesting a climate model at this time.

When the debts/resource-share are more in favour of running a climate model, it will request one.

OK, sorry to have bothered you guys, but I always ran with \"no new tasks\" selected and am pretty sure I got that \"not requesting new work or reporting completed tasks\" message every time I contacted the server for any reason (except of course when reporting a complete model).

---

Visit the Scotland team

The words are similar, but their meaning changes, depending on the \"tab\" in which they appear.

In the Projects tab, they are talking about whether or not you have permission to get new work.
In the Messages tab, they\'re talking about whether or not you want to get new work.

A bit like standing at a railway level crossing.
If the gates are closed, lights flashing, and a bell ringing, you don\'t have permission to cross.
But when the lights and bell stop, and the gates open, you don\'t HAVE to cross; you may just be standing there train spotting.

Grizz has had 4 climate models

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/results.php?hostid=520909

of which the last crashed at the end of September. I find it difficult to believe that CPDN is still in debt to the other projects four months later. Unless he ran the model continuously and suspended his other projects all the time the model was running.

Grizz, here are a couple of ideas to try. Before you try them, set all your other projects to No New Tasks. Then in your BOINC manager advanced preferences, lengthen the work buffer to the max of 10 days.

* Reset the CPDN project. Because you have no running model, this will do no harm.

* If that doesn\'t work, you could detach from CPDN then reattach.

---

Cpdn news

Les is concerned about the \'No more tasks\' option, which is the most likely cause for your problem.

But there is a second possibility which is that the Boinc manager is paying back \'debt\', - what are your resource shares for the climate project compared to other projects?

The reason that I ask is that people with many projects often run into an issue where the climate model gets priority for a while over the other projects due to the low resource share putting the model\'s deadline at risk. As a result, the model gets much more than its fair share of time. Once the model has finished, the other projects get \'paid back\', by being given priority in return. This can be for quite an extended period of time.

My situation is a little bit different, but similar in that no work is being done for cpdn lately. I just began a few days ago, so of course I have not completed or uploaded any tasks, with just 20% of resources of an Athlon 2200+. After about 21 hours of work on cpdn, and suspending all 5 of my projects to upload to Einstein@home when it got confused, cpdn\'s Status is now \"Waiting to run\" in Tasks tab. Is this likely also because of this \'debt\' variable? If so, is there a user-configurable switch, to ignore debt? I\'d rather just let all projects use as much of the 20% available to them as they can, without reference to past resource allocation, whatsoever.

Other detail: I have only one cpdn task, and it\'s due next January, which is why I guessed that some resource debt algorithm was classifying that low priority. Sorry if this is all off-topic.

The more you fiddle, the more problems you create.
BOINC was designed to look after these matters on a long term basis, not hour by hour. Leave it to get on with things.
Running the extremely long cpdn models alongside the very short work units from other projects has always caused BOINC to panic the first time it gets a climate model, but it will sort things out in the long run.

And there is NO deadline used in this project. The 1 year figure is just there because some number HAS to be.

\"Waiting to run\" just means that another task is taking it\'s turn on the CPU. For example, if you have three tasks, and a dual-core processor, two will be \"Running\" and the third will be \"Waiting to run\". They\'ll swap around as needed.

As Les says, the best thing to do is just to let Boinc \'Do it\'s thing\'.

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

Boinc DV is safe, use that to reset debt to zero, when ever you want to just exit Boinc first.
Regards
Masud.

\"Waiting to run\" just means that another task is taking it\'s turn on the CPU. For example, if you have three tasks, and a dual-core processor, two will be \"Running\" and the third will be \"Waiting to run\". They\'ll swap around as needed.

As Les says, the best thing to do is just to let Boinc \'Do it\'s thing\'.

I see, thanks for the explanation.

Not necessarily, if you click \'update\' but the manager doesn\'t want to download more work, you\'ll get that message. The tricky thing is to try to work out why...

I am using dial-up, so I have my dialer set to ask me to connect. I use this 2nd line for Internet, FAX and as a 2nd voice line (grandson!).

On 21 Feb, 2008 CPDN tried to connect to get new work (log msg indicated this), but since it was unable to connect, of course the log indicated the \"http://... failed.\"

When I do go online and click on \"Update\" for CPDN, BOINC sends the msg:

3/18/2008 12:02:53 AM|climateprediction.net|Sending scheduler request: Requested by user
3/18/2008 12:02:53 AM|climateprediction.net|(not requesting new work or reporting completed tasks)
3/18/2008 12:02:56 AM|climateprediction.net|Scheduler RPC succeeded [server version 509]

This happens each time \"I\" send the scheduler request. But if BOINC sends the schedule request (which is doomed to not connect), it DOES request more work!

I am also running SETI and Einstein and those two get, complete, report & get new work constantly. If this continues, I\'ll just have to change the dialer\'s setting to allow BOINC to dial-up the Internet and just hope it doesn\'t interfere with a FAX, etc.

---

Hi,

I think that most likely Boinc is paying back the other two projects for extra time that CPDN has received in the past. So for a period it will not get any work for CPDN, but once this time is up, it will ask for work. Boinc calls this \'long term debt\' (i.e., one project owes CPU time to another project).

-- Edit:

I\'ve just seen a new way to modify this \'long term debt\' figure.

To use this, you need a recent version of the Boinc manager, and then you\'ll need to open a \'command prompt\' (start/programs/accessories), navigate to the Boinc directory (type cd \\program files\\boinc), and then type (for example):

boinc_cmd --set_debts http://climateprediction.net/ 0 10000

The 0 is the new \'short term debt\' figure which is used in deciding which project gets the next hour / 3 hours etc of CPU time. The 10000 is the new \'long term debt\' which is used to decide which project(s) need to download new work from the servers. If I recall correctly, negative numbers are where the project owes time to other projects, and positive numbers are when the project is owed time from other projects. It is in seconds, so 10,000 is slightly over 3 hours.

I haven\'t tried this myself.

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ