Rosetta not wanting to crunch on my Linux Box

Message boards : Number crunching : Rosetta not wanting to crunch on my Linux Box

To post messages, you must log in.

AuthorMessage
Profile Dragokatzov

Send message
Joined: 5 Oct 05
Posts: 25
Credit: 2,446,376
RAC: 0
Message 37913 - Posted: 17 Mar 2007, 14:53:50 UTC

And its not the first time I have had this problem either... I am running Ubuntu 6.06 and Rosetta worked for 1 or 2 days, then decided to be a pain in my but and not work. its been stuck on "waiting to run" or a long time. I also have WCG running on this computer as well, and it runs well, sometimes i only have this one Rosetta work unit "waiting to run" and no WCG WU's. on this system i have an Athlon XP 1700+, and 512 megs og ram, so it s not like theres not enough system resources avaliable. anyone have any ideas?
Victory is the ONLY option!
ID: 37913 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Thomas Leibold

Send message
Joined: 30 Jul 06
Posts: 55
Credit: 19,627,164
RAC: 0
Message 37926 - Posted: 17 Mar 2007, 18:35:36 UTC - in response to Message 37913.  
Last modified: 17 Mar 2007, 18:37:58 UTC

on this system i have an Athlon XP 1700+, and 512 megs og ram, so it s not like theres not enough system resources avaliable. anyone have any ideas?

For some Rosetta workunits (the recent HINGE series) 512MB is the minimum, but most workunits use much less.


The issue you are reporting is a longstanding and still not resolved issue with Rosetta. The workaround is to change the BOINC General Preferences and enable (set to "Yes") the option:

"Leave applications in memory while suspended?
(suspended applications will consume swap space if 'yes')"

This is always required when using multiple projects since the Rosetta linux client doesn't handle the rescheduling properly unless it is permitted to remain in memory (it fails with segmentation violation but doesn't terminate properly).
Team Helix
ID: 37926 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Dragokatzov

Send message
Joined: 5 Oct 05
Posts: 25
Credit: 2,446,376
RAC: 0
Message 38142 - Posted: 22 Mar 2007, 20:08:07 UTC - in response to Message 37926.  

hey i tried it and it works like a charm! thanks a lot!
Victory is the ONLY option!
ID: 38142 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Milkman

Send message
Joined: 23 Nov 05
Posts: 2
Credit: 5,497,447
RAC: 0
Message 38850 - Posted: 2 Apr 2007, 14:25:52 UTC
Last modified: 2 Apr 2007, 14:26:34 UTC

The issue you are reporting is a longstanding and still not resolved issue with Rosetta. The workaround is to change the BOINC General Preferences and enable (set to "Yes") the option:

"Leave applications in memory while suspended?
(suspended applications will consume swap space if 'yes')"

This is always required when using multiple projects since the Rosetta linux client doesn't handle the rescheduling properly unless it is permitted to remain in memory (it fails with segmentation violation but doesn't terminate properly).


Perhaps I am just not good at searching forums for answers, but I have been trying to resolve the "dead" Rosetta WU on my Linux machines for some time and couldn't find this answer until today. Don't know that it will solve my problems, but I suspect it will (the description fits my symptoms). Should this be mentioned in the FAQ under a Linux heading or some such as it has led to many lost days of computation over the months? Perhaps it is there and I just couldn't find it, but my symptoms are (hopefully now were):

CPU usage would drop to nothing, open BOINC manager and find that a Rosetta unit was loaded and "running" but nothing was happening. Other clients (SETI and climateprediction) would not switch in, so the machine would just sit idle until Rosetta units aborted. Perhaps a client restart would have fixed the WUs, I never thought to try that.
ID: 38850 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 38856 - Posted: 2 Apr 2007, 16:25:56 UTC - in response to Message 38850.  

... Should this be mentioned in the FAQ under a Linux heading or some such as it has led to many lost days of computation over the months?


I keep hoping a new version of BOINC will correct the problem. BOINC is responsible for managing which task is running and when. I've been working on improvements to the FAQ. To date, have only had time for things that are the same on all platforms and that will not likely change in the near future.

I suppose I should devise a troubleshooting section, which would apply to all platforms. And then include ending and restarting BOINC at some point in those steps. I will keep your suggestion in mind when I get back to working further on the FAQs.

I agree, with subject lines like "not wanting to crunch", it makes it difficult to find adequate search words.
Rosetta Moderator: Mod.Sense
ID: 38856 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Thomas Leibold

Send message
Joined: 30 Jul 06
Posts: 55
Credit: 19,627,164
RAC: 0
Message 38913 - Posted: 3 Apr 2007, 6:04:14 UTC - in response to Message 38850.  

Should this be mentioned in the FAQ under a Linux heading or some such as it has led to many lost days of computation over the months? Perhaps it is there and I just couldn't find it


The recommendation for the preference setting can be found under "Additional Notes" in the Rosetta System Requirements. I doesn't say why, but there are a number of threads where this topic is raised in both "Questions and Answers" and "Message Boards".

Instead of documenting the workaround I would prefer to see this problem fixed.

Given that this is a Rosetta specific problem (other projects have no such issues when task switching) I'm not sure that waiting for new versions of Boinc is going to magically fix the problem. To be clear, I'm not saying that the cause of the problem is necessarily in Rosetta code. It could be a unique way of using Boinc APIs that triggers a problem inside Boinc code. However as long as this issue only effects Rosetta, why would any Boinc developer search for the cause in their code (if they even heard about it) ? Right or wrong, I'm sure that they will assume it is something in the Rosetta code that is causing it.

Team Helix
ID: 38913 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : Rosetta not wanting to crunch on my Linux Box



©2024 University of Washington
https://www.bakerlab.org