I have a Linux-x64-Client of boinc 8.2.5 on an Ubuntu 24.04.x

The client doesn't have any WU to crunch, it should fetch work from CPDN.

But instead of doing this, the client tells in Logs:

Manni HyV Li

884 climateprediction.net 22-08-2025 12:54 update requested by user
885 climateprediction.net 22-08-2025 12:54 Sending scheduler request: Requested by user.
886 climateprediction.net 22-08-2025 12:54 Not requesting tasks: don't need ()
887 climateprediction.net 22-08-2025 12:54 Scheduler request completed
888 climateprediction.net 22-08-2025 12:54 Project requested delay of 606 seconds

CPDN has a ResourceShare of 100, all other projects are on "no new work"

The client was a fresh install on a "clean" machine, so no more history can influence the client

Any idea what to do ?

---

Currently the Server status: Tasks ready to send 0

I think only Windows 64 bit with high memory and disk requirements have been sent out in last few weeks.

Shure

The client should ask the server for work like this client:

WoolyW10 VLneu

749 climateprediction.net 21-08-2025 23:18 Sending scheduler request: To fetch work.
750 climateprediction.net 21-08-2025 23:18 Requesting new tasks for CPU
751 climateprediction.net 21-08-2025 23:18 Scheduler request completed: got 0 new tasks
752 climateprediction.net 21-08-2025 23:18 Project has no tasks available
753 climateprediction.net 21-08-2025 23:18 Project requested delay of 606 seconds

Then the server can decide weather it wants to hand out some WUs or not, but if the client as shown in my first post doesn't ask for work the server has no need to hand out work

Actually it's only Linux tasks that have been sent out. Of course if you want to run them "under Windows" one will have to use some sort of virtualisation.....

Memory & disk requirements are substantial, something like 26GB per task, plus a lot of disk space....

Ubuntu 24.04

The machine has 32 GB memory.

Normal tasks from CPDN use something around 2 GB per Task, only the new MultiCore-App uses up to 26 GB

But the client doesn't know all this. He would have to request new work and then the server decides weather he will handout work to the client or not.

Right now, main.cpdn.org is down but since some time yesterday there hasn't been any work of any sort bar the occasional resend that has failed on another machine.

Once Glenn sorts out the problems happening with the testing batches, the next batch will be the high memory multi-core ones.

Edit: If you have had tasks crashing, you will be limited to one per day till you have returned a successful task. This can have unfortunate side effects when testing batches go out on the main site.
This is a more likely explanation of what is happening than something being wrong with the client.

Will it still be linux only for that new app in CPDN ? any plan to go to other OS... like macOS ? :)

In reply to JeromeC's message of 22 Aug 2025:

Will it still be linux only for that new app in CPDN ? any plan to go to other OS... like macOS ? :)

Glenn who has done the programming for the OIFS tasks has been clear he doesn't have the time to do the work to extend them for other platforms. For quite a long time all work at CPDN has been single platform so that for each task type only one platform needs support. That means for the Mac, recent machines are no good as the older met office models are all 32 bit.

CDPN has had no available work for several days but I've also seen the client act pretty dumb with zero tasks running and not download work. It can sometimes take several attempts.

CDPN has had no available work for several days but I've also seen the client act pretty dumb with zero tasks running and not download work. It can sometimes take several attempts.

Does this work fetch issue as previously described show up on any other projects you are attached to on this computer when work is available?

If so, please post details here and hopefully one of the BOINC Developers will chime in.

In reply to Dr Who Fan's message of 23 Aug 2025:

CDPN has had no available work for several days but I've also seen the client act pretty dumb with zero tasks running and not download work. It can sometimes take several attempts.

Does this work fetch issue as previously described show up on any other projects you are attached to on this computer when work is available?

If so, please post details here and hopefully one of the BOINC Developers will chime in.

I have seen this occasionally on different clients from me, starting with BOINC 8.x.x and on different projects, mostly with starting Formular-Boinc-Sprints. Then I never had the time to follow up this problem

In reply to Yeti's message of 24 Aug 2025:

In reply to Dr Who Fan's message of 23 Aug 2025:

CDPN has had no available work for several days but I've also seen the client act pretty dumb with zero tasks running and not download work. It can sometimes take several attempts.

Does this work fetch issue as previously described show up on any other projects you are attached to on this computer when work is available?

If so, please post details here and hopefully one of the BOINC Developers will chime in.

I have seen this occasionally on different clients from me, starting with BOINC 8.x.x and on different projects, mostly with starting Formular-Boinc-Sprints. Then I never had the time to follow up this problem

This is pretty much my situation too with suspending work to be under my work buffer limit or even no work buffer. The client would still need multiple communication attempts with another project server to actually download work. It's like the client is still working off an old status and won't tell the new project to send work. Doing this with CDPN is just more of a pia due to the 10min delay between updates.

It looks as if it has something to do with an app_config.xml.

I just removed it and now the client asks for work

Not a good idea to run CPDN without an app_config.xml

Doing this with CDPN is just more of a pia due to the 10min delay between updates.

Till Glenn changed it recently the backoff time was an hour.

Not a good idea to run CPDN without an app_config.xml

I have been running CPDN without an app_config.xml for many years without it causing problems. Perhaps it only becomes an issue when you are running other project(s) at the same time. When CPDN has work, I don't crunch for any other projects. (I may put in an app_config file though for managing the multi-core tasks alongside others.