LHC@home

With the new Vbox version and extension pack I run into headless errors with this current batch of tasks. Not always but randomly.

But I think, can't be sure, that when I was having trouble getting work for ATLAS I set my tasks allowed to 2 from 1 for CMS and Theory. But now that I am getting primarily ATLAS tasks again, I set tasks allowed back to 1.

Here are a few links to the "bad" ATLAS tasks

https://lhcathome.cern.ch/lhcathome/result.php?resultid=375188194
https://lhcathome.cern.ch/lhcathome/result.php?resultid=375147897 - this ran for an hour
[/url]https://lhcathome.cern.ch/lhcathome/result.php?resultid=375147897[url] - ran one minute and died

Common error code is 2147467259 (0x80004005)
I see at the bottom of the stderr text something about Vbox could not start....

Vbox is clean with only the current task running.
When I shut down for the night, I suspend the tasks, shut down BOINC client and exit the program.

Am I correct with my theory about running more than 1 tasks of ATLAS is causing these problems?
It's been a long time since I have had any problems.

Win10 fully updated computer.
AMD Ryzen 7 3700x - 8 physical 8 virtual cores.
15 in use 1 reserved for windows.

Atlas need 24/7.
When you save the task and start it again next morning, they not starting well.
Virtualbox is not the reason.

I've been doing ATLAS this way for years.
Only time I have had this problem is trying to run 2 at the same time.
I think that is the cause, just double checking.

Maybe, one Atlas Task is running well after restarting in the morning,
the 200 collisions beginning again by one.
When Windows need a restart, have to ending all Atlas Tasks, before doing it.

well again, due to the way the system processes these tasks, it is a one by one task. Finish one, download a new, repeat.

The problem probably is, when you run two or more and you shutdown your machine BOINC wants to write those tasks to disk within 30 seconds, but can't.
When restarting those tasks the next day one or more tasks are not written correctly to disk and can't restart.

The problem probably is, when you run two or more and you shutdown your machine BOINC wants to write those tasks to disk within 30 seconds, but can't.
When restarting those tasks the next day one or more tasks are not written correctly to disk and can't restart.

Ah ha! That makes more sense.
But also long ago, for some reason, as you might recall, I could not run 2 tasks at the same time continuously without one or both erroring out which is why I switched to 1 by 1.
That seems to have gone away, but then what you describe happens instead.
So for me 1 by 1 is a safe bet with no problems.

Thanks for the explanation.

Just found this in your logfiles: Setting CPU throttle for VM. (97%)

I have already switches to Atlas-Native, but I remember that in the past CPU-Throttling should be switched off with VB and ATLAS

---

Supporting BOINC, a great concept !