Posts by Quez

1) Message boards : Questions and problems : Constantly running 1 WU of each selected project (Message 79595)
Posted 13 Jul 2017 by Quez
Post:
Dear Ageless,

running 20+ (Windows) VMs produces a massive overhead of CPU and RAM load. I now solved it by running a .cmd script opening enough --allow_multiple_clients clients to run every project.

Thank you all very much for the input!

Kind Regards, Quez
2) Message boards : Questions and problems : Constantly running 1 WU of each selected project (Message 79573)
Posted 13 Jul 2017 by Quez
Post:
Thank you very much, I will try BOINC 6 . And I messed around with --allow_multiple_clients but it seems the different clients are interfering with each other sometimes, especially when using BAM. Once I disable a project on one client, it becomes disabled on another, too, where it is supposed to run. But maybe I just messed up everything with 10+ clients running.

When you say BOINC 7 is intermixing just when switching between projects, can I just set the "time between switching" to 9999999 or similar? I never fully understood this setting. Or is it just switching tasks within one project?
3) Message boards : Questions and problems : Constantly running 1 WU of each selected project (Message 79549)
Posted 12 Jul 2017 by Quez
Post:
Well, how can I do it automatically? Or which command do I need? I am not worried of idle threads.

In any case, thanks for the response.
4) Message boards : Questions and problems : Constantly running 1 WU of each selected project (Message 79545)
Posted 12 Jul 2017 by Quez
Post:
So it is not possible to run the same number of tasks in several projects, to restrict the number of workunits a project loads, or to ignore the "cache is full" statement if a new task could be run then?
5) Message boards : Questions and problems : Constantly running 1 WU of each selected project (Message 79543)
Posted 12 Jul 2017 by Quez
Post:
Dear BOINC community,

I am trying to run 1 WU of several projects on my brand new server. E.g. I always want to have 1 LHC WU and 1 Universe WU running, but with more projects - CPU only.

My problem is that the cache runs full at some point so that projects can't download work anymore.

I have set <project_max_concurrent> to 1 in each project.
I tried setting max work and cached work to 0, but the issue persists although I read a comment where Ageless (I think) indicated that BOINC will then always download as much work as necessary to max out the hardware.
--fetch_minimal_work is not what I am looking for as I want to run it constantly

I am trying to avoid --fetch_minimal_work with a client restart every time a WU gets finished, which could be a workaround I suppose. I am alos trying to avoid setting up several virtual boxes for every project I want to crunch.

Is there any possibility to run the same amount of workunits? Or to restrict the number of workunits a project can cache? Or let BOINC ignore that the cache runs full IF a project could run one more task?

Kind Regards, Quez



Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.