LHC@home

A fedora 36 machine will run ATLAS jobs fine. But sofar all CMS jobs fail with
"00:00:05.652853 Console: VM runtime error: fatal=false, errorID=VMBootFail message="The VM failed to boot. This is possibly caused by not having an operating system installed or a misconfigured boot order. Maybe picking a guest OS install DVD will resolve the situation".

VBox works well with Atlas but not with CMS. Vbox ver 7.06.

Any clues?? other than dont run cms.
Ta
Nairb

This log entry points out that the VirtualBox disk entries (CMS only) are messed.
This happened earlier but can't be resolved by vboxwrapper.

VBoxManage -q closemedium "/home/nairb/projects/lhcathome.cern.ch_lhcathome/CMS_2022_09_07_prod.vdi" 
Output:
VBoxManage: error: Cannot close medium '/home/nairb/projects/lhcathome.cern.ch_lhcathome/CMS_2022_09_07_prod.vdi' because it has 1 child media
VBoxManage: error: Details: code VBOX_E_OBJECT_IN_USE (0x80bb000c), component MediumWrap, interface IMedium, callee nsISupports
VBoxManage: error: Context: "Close()" at line 1878 of file VBoxManageDisk.cpp

ToDo:
1. Stop BOINC
2. Start VirtualBox GUI -> Media Manager
3. Remove all entries related to 'CMS_2022_09_07_prod.vdi' (children & parent). Do NOT remove the file itself!
4. Leave VirtualBox GUI and restart BOINC


Other hints:
The computer reports only 12 GB RAM.
This is fine as long as there is enough RAM headroom for the OS and it's page cache.
Check regularly how much swap is used. A few 100 MB would be OK, a few GB would not be OK, the more the worse.
And, no, there's no hard limit that tells you when it is too much.

Many thanks for the info.
I followed your instructions and now the machine is running 2 cms jobs fine. It's only using half the available ram, so might even be able to run 3 jobs at once,
I have set the preferences to have only 2 jobs at once to be sure.

All is now running fine

Ta
Nairb