LHC@home

Dear Friends,

An updated* mcplots codes is subscribed for the BOINC production now.

This time the update addressed Sherpa event generator and particularly issues with the endless loops which should be significantly reduced or disappear now.

Happy crunching!
The Team

*: http://mcplots-dev.cern.ch/production.php?view=revision&rev=2390

51min average time......will this be the same with the Theory-dev version?

(since I have one running over 12 hours right now)

This time the update addressed Sherpa event generator and particularly issues with the endless loops which should be significantly reduced or disappear now.

Thanks Anton,
I can't wait to get tasks from revision 2390 (I've 2378-tasks queued to crunch away first).

This time the update addressed Sherpa event generator and particularly issues with the endless loops which should be significantly reduced or disappear now.

Thanks Anton,
I can't wait to get tasks from revision 2390 (I've 2378-tasks queued to crunch away first).

I'm downloading some of each (2/3rds the old version). Most work, but one has been running 3 days 9 hours. I changed it to a 40 day limit to see what happens. If it goes past the deadline, is there any point in continuing it? Does LHC still want it back if it takes a little longer?

---

With the new version I got several Sherpa's and they all were Valids and all the different event generators ran some tasks that actually were around 18hours each instead of lots of the 1 hour ones that make you watch them all the time (well I do that anyway)

But now I have seen some Pythia's that wanted to run 10 days to get that Invalid aka computer error ....so they get Aborted when caught in the act.

I have a Pythia now that is at 75 hours and the log says it is running normal but I don't trust it so later when I check and it is still running it gets Aborted.
(unlike those tasks that start at Failed but keep running anyway) usually Aborted in the first 3 minutes. but I am running the 5.20 Version of Theory but decided to d/l the 300.05 vdi Version here to see how that goes (since my ip is running fast for some strange reason)

Error during processing histograms. https://lhcathome.cern.ch/lhcathome/result.php?resultid=272880928

I got one that claimed to still be running after over a day, yet there was 0.1% CPU usage, 0.1MB/sec disk activity, and no internet activity. I aborted it.

---

I had one similar (boinc pp jets 7000 40,-,560 - sherpa 1.4.3 default 100000 0). No progress on ALT-F2, but 'top' shows sherpa 98% busy. The task came from LHC-dev, but fishing in the same pool.
Coincidence that the last line was dumping histograms or did the VM not recovered properly after a suspend and resume?
Task: https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=2899842
The job should be ready yesterday night, but I suspended the task before and resumed the next morning.

I had one similar (boinc pp jets 7000 40,-,560 - sherpa 1.4.3 default 100000 0). No progress on ALT-F2, but 'top' shows sherpa 98% busy. The task came from LHC-dev, but fishing in the same pool.
Coincidence that the last line was dumping histograms or did the VM not recovered properly after a suspend and resume?
Task: https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=2899842
The job should be ready yesterday night, but I suspended the task before and resumed the next morning.

Suspending is a real problem for Virtualbox stuff. I've therefore set all my computers to "switch tasks every" 1000000 minutes - which I think is a few hundred years. So they'll always finish what they started before going onto something else. Seems a more sensible way than constantly changing tasks around anyway - I see no point in doing a quarter of this task, then a quarter of another, then swapping back and forth and messing around.

If I shut down or reboot a Windows computer, Virtualbox always has "open connections" or something. I often just click "shutdown anyway", maybe I shouldn't. If I leave it, after 5 minutes Windows gives up and goes back to the desktop without shutting down. If I try again, there's no complaint, so I guess it's just very slow at stopping.

Stuck LHC tasks usually seem to happen if I've been farting around with the computer. At the moment a lack of RAM is really annoying Boinc, and even with the "switch tasks every" setting I mentioned above, it's still swapping back and forth as it changes its mind about how much RAM is free. The UK mail system is pathetically slow at the moment for no apparent reason (private couriers are operating normally), so I'm having to wait for extra RAM chips....

---

Hi Crystal,
This is an error in Rivet package which we use for analysis and histogramming, I have notified authors. It will be fixed in the some of the next version and the error will disappear.
Thanks for the pointer!

http://mcplots-dev.cern.ch/production.php?view=status&plots=hourly#plots
Is not running ftm.

McPlot is back since yesterday, thank you.
Is there a log to fill the Data
from 20/12/28-21/01/05 AND 21/01/22-21/01/25
2020-12-27 17051 328 17379
2020-12-28 253 3 256
2021-01-05 42819 768 43587

2021-01-21 3644 90 3734
2021-01-22 1758 51 1809
2021-01-25 8022 139 8161
or is this Data included?