Bad Results from LHC@home
Hi everyone,
In the last 10 days I noticed several errors from my tasks in the LHC@home project.
Out of the total 952 tasks, only 347 are valid. One finished in a "Validation inconclusive" state, while 542 tasks were ended with various errors.
I analyzed the different error cases and I found several categories:
- All ATLAS Simulation work units are having issues with the VirtualBox on my 64-bit Windows 10 laptop: "Error creating VirtualBox instance! rc = 0x80004002". I updated the VM VirtualBox to version 5.2.12, and the situation is persistent. For the time being, I decided to remove ATLAS Simulation from my settings.
- Hundreds of SixTrack work units (421 tasks) from a specific machine are finishing with zero execution time and a "Process got signal 11" error message. This is an experimental virtual core on Google Cloud, configured with the minimum 0.6 GB of memory. Does anyone have a clue about this strange signal 11?
- About twenty tasks were canceled immediately by the server.
- About a hundred work units stopped after about 20 minutes of CPU time with the "Disk usage limit exceeded" error on a second virtual machine in Google Cloud (8 vCPUs with 7.2 GB of memory).
I am concerned about the high number of fails from LHC@home. Is it only specific to my machines, or a more generic topic?
Right now, for CPU only i personally get the best results on Cosmology and VGTU.
I was crunching LHC@home as well for a while without problems, and then started getting "computational error" problems back in the winter for every one of my LHC work units. I upgraded VirtualBox and had no success. I couldn't figure out what went wrong so I started crunching other projects with the plan of eventually returning and figuring out the problem.