Rosetta@Home with BOINC & Linux (Ubuntu)

[Robor]

Came home tonight to find my 'Server' system running but sitting idle. I want to help but I don't have the time or patience to babysit or troubleshoot it - especially when this system was only on for that purpose. Maybe BOINC/Rosetta will get a stable Linux client and I'll come back but until then I'm going to have to go dark. Sorry TeAm...

=================

I'm having a problem with one of my Rosetta@Home clients stalling. I'm new to R@H so I'm not sure where to start troubleshooting. I normally launch the the client from the command line (sudo /etc/init.d/boinc-client start). I can monitor it via a 'top' command and see if it's using CPU. It will fire off and work for a while then for some reason just stop. I looked in the BOINC Manager and saw there's a 'Messages' tab with a log but I already did a stop/start on the client so I don't think I have anything containing the previous log - I searched for a 'boinc log' in the file system but didn't see one. Any suggestions on where to start the next time the client stalls? Here's my current log...

 

[BlackMountainCow]

What kind of CPU is that? I see from the log that you were already past the deadline of that WU when you finished it. What's the value of your "connect to network every X days" in the Rosetta site settings?

 

[Robor]

That's my work laptop so I can't look at it right now. I know it's a Centrino system so a Pentium M I think?

Where should I be looking for this setting? If it is this...

Connect to network about every
(determines size of work cache; maximum 10 days)

...it is set at 2 Days

I think my problem is my client is bombing out for some reason. All of my systems run 24/7. I have 2 other Linux machines that run under the same config but no stalls like I'm seeing.

 

[BlackMountainCow]

Yes, that's the setting I was looking for. Looks all right.

From your log, it just seems as if BOINC stops after finishing a WU. No reason why, though. I have no Linux PCs, so I can't help you on that one. Did you search the Rosey forums already or post there?

 

[petrusbroder]

Looked at the log.
I did not see any "error" message either. Could thus be a "new" bug in the application from rosetta or some BOINC-setting.

Is the log you have posted a complete log - i.e did the message log just end as shown? If yes, that is somewhat unusual ...

Stupid question: do you see the rosetta process running (i.e. using CPU resourses) or is it just sitting there doing nothing? have you tried - in the BOINC-manager - to reset the project? You can do that on the projects tab ... or you can try to disconnect from the project and then to reconnect again.

Please post if it does or does not resolve the problem ...

 

[petrusbroder]

Originally posted by: BlackMountainCow
Yes, that's the setting I was looking for. Looks all right.

From your log, it just seems as if BOINC stops after finishing a WU. No reason why, though. I have no Linux PCs, so I can't help you on that one. Did you search the Rosey forums already or post there?

As I read the log BOINC-mananger starts a task:

Wed 11 Apr 2007 02:56:12 PM EDT|rosetta@home|Starting task BENCH_04JUMPING_SAVE_ALL_OUT_-1bm8_-_NATIVE_PAIR_32_74_BARCODE_R20L_filters_161
_1201_2 using rosetta version 554

but it never exits that task; it does not end it either.
So we see a task start, and then stop without leaving a trace. Normally an error message would appear, such as "computation error", "task exited without end-of-task message" or similar. An exit without an error message is unusual.
That's why I am asking if the rosetta-application is still memory resident, using CPU-resources or if it has exited and if the log is complete ...

 

[Robor]

Okay I looked in the BOINC manager (GUI) and saw under the 'Tasks' tab that all of my stuff had a report deadline of April 8th. I aborted all of them then detached and reattached. It's downloading more work right now and that has a deadline of April 22. As I'm typing this it finished the download and Rosetta is using CPU time again. Will report back if I continue to have issues. The good news is my other systems are a-okay so far (I know - just jinxed 'em).

 

[petrusbroder]

 

[BlackMountainCow]

 

[Robor]

Update

 

[BlackMountainCow]

Crap!