Task completion time - Wildly different - Expected?

Everything about the project RNA World
Nachricht
Autor
Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Task completion time - Wildly different - Expected?

#1 Ungelesener Beitrag von Jacob Klein » 02.04.2014 04:20

I am working on a the following work unit:
http://www.rnaworld.de/rnaworld/workuni ... id=6330945

I am computer 16705. BOINC says my VM task has been going for 1,105 hours (~4,000,000 secs), progress is around 41%, the VBoxHeadless process still uses CPU, and the VM still regularly checkpoints every 10 minutes.

But it would appear that my wingman completed this task in 587,918.81 secs.

- Is this expected?
- Can his/her result possibly be valid?
- Is my task somehow stuck?
- Is there any way to know if my VM task is progressing, other than to check BOINC progress and to monitor CPU usage of VBoxHeadless?
- Should I keep running this task?

Regards,
Jacob Klein

Benutzeravatar
yoyo
Vereinsvorstand
Vereinsvorstand
Beiträge: 8170
Registriert: 17.12.2002 14:09
Wohnort: Berlin

Re: Task completion time - Wildly different - Expected?

#2 Ungelesener Beitrag von yoyo » 02.04.2014 08:17

Your wingman produced a result which is (compressed) only 384 bytes and a log which is (compressed) only 907 bytes. I assume that this is not correct.
HILF mit im Rechenkraft-WiKi, dies gibts zu tun.
Wiki - FAQ - Verein - Chat

Bild Bild

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Task completion time - Wildly different - Expected?

#3 Ungelesener Beitrag von Jacob Klein » 02.04.2014 11:45

So, just to confirm, I should continue to run it, and the validator should require a 3rd result result later to validate my result. Correct?

ChristianB
Admin
Admin
Beiträge: 1920
Registriert: 23.02.2010 22:12

Re: Task completion time - Wildly different - Expected?

#4 Ungelesener Beitrag von ChristianB » 02.04.2014 11:55

Yes, this is correct. Validation is only executed when the quorum is reached. It would be a nice addition to sanity check the result files upon arrival. I will mention this to David.

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Task completion time - Wildly different - Expected?

#5 Ungelesener Beitrag von Jacob Klein » 02.04.2014 12:07

Thank you both!

This is my first "big" VM task that has been running in an environment that is stable enough to hopefully complete, so you can understand why I wanted to make sure things are going correctly. With your answers, I feel more confident that this one could be my crowning success from all the VirtualBox/vboxwrapper testing that I had done previously :)

ChristianB
Admin
Admin
Beiträge: 1920
Registriert: 23.02.2010 22:12

Re: Task completion time - Wildly different - Expected?

#6 Ungelesener Beitrag von ChristianB » 02.04.2014 12:32

and you fully deserve this ;)

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Task completion time - Wildly different - Expected?

#7 Ungelesener Beitrag von Jacob Klein » 25.04.2014 04:28

... that task is now approaching 1600 hours, and still going strong. Loving these RNA World VM tasks (I'm running 2 of them on the same PC)!

Jacob Klein
Brain-Bug
Brain-Bug
Beiträge: 564
Registriert: 26.07.2013 15:41

Re: Task completion time - Wildly different - Expected?

#8 Ungelesener Beitrag von Jacob Klein » 03.06.2014 18:02

The original task is still crunching, now at the 2400-hour mark, with progress of 89.949%. The 2nd task that I'm also running on the same machine, is at the 1700-hour mark, with progress at 59.161%. So far so good, I think, for both tasks.

As a precaution, every 5 days, I shutdown BOINC, and go into the Oracle VM VirtualBox Manager, and do a full-clone (selecting option "everything", which includes the snapshots), for each of the VMs. It may take up tons of disk space, but this way, even if the task has an error, or completes but then doesn't get reported correctly, I'll have cloned VMs where we could finish them manually if need be. :) I just don't want to see 2400+1700=4100 hours of work disappear.

I'll keep crunching. Looking forward to seeing one of these complete! Don't care at all about the credit - just want to see a successful finish!

Zurück zu „RNA World Discussions (english)“