Orbit@Home is alive!

[NeoGen]

The orbit@home code development has resumed since the beginning of March 2010. We are now writing and testing the programs that assimilate observational data and prepare work units. Next we will complete and test the BOINC application. Initial test WUs should be released in about 2-3 weeks.
Technically inclined users can follow the details of the coding progress, such as the summary or the specific BOINC application tree codenamed SurveyReview.

I got one of these SurveyReview workunits for windows. These will be my very first points on this project! WOOHOO!!
It appears to take between 2 and 3 hours to crunch, and doesn't have graphics/screensaver capability.

[BobCat13]

They don't appear to checkpoint either, at least nothing in the boinc message log about checkpoints.

[NeoGen]

I have gotten a hand full of them already... they keep coming one by one. Might be a server side setting to not let them cache.

I got the finished ones all still pending... come on guys, I need some wingmen here. Let's get the Team up to the Top 20 again!

[velociraptor]

wingman velociraptor ready for duty.
never thought i will see this project again...
sometimes it is really nice when reality proves me wrong.

[AMDave]

2-3 hours?

Thu 20 May 2010 18:46:54 EST orbit@home Starting task LINEAR_2009321_3962_1274314638_0 using SurveyReview version 105
Thu 20 May 2010 18:46:57 EST orbit@home Computation for task LINEAR_2009321_3962_1274314638_0 finished

I got seconds.
Better do some more to see where this goes...

[AMDave]

It goes nowhere so far:

Thu 20 May 2010 18:56:43 EST orbit@home Scheduler request failed: HTTP internal server error

lots of these in a row.

[AMDave]

getting some more now, short wu's (2 seconds to 70 seconds), small downloads. Monitoring.

/ed -
moments later some longer wu's arrived. looked short, ran to 100% then went back to 1 or 2 %

current project thread contains similar and additional observations
http://orbit.psi.edu/oah/forum_thread.php?id=367

another observation I had was that BOINC 6.10.45 suddenly got really confused about what was running
nothing else (non-BOINC) was running on the machine but the BM suspended all tasks because there was too much CPU useage
I have that set at 75% non-BOINC CPU activity
then it obviously sampled again a few seconds later and saw that it was wrong and resumed
then it happened again a couple of times
now it seems to have settled down.

I think it was the project-switching that caused it.
When BOINC issues a suspend to a project task, it mostly assumes that the suspend takes effect immediately.
That is not the case for some projects which may have several threads in flight on a single core and can take up to a couple of minutes to get to the next interrupt

Thus I think BOINC looks at the CPU useage and sees that the CPU is still in use,
sees that something that it does not think belongs to it is hogging the CPU,
hits the non-BOINC CPU limit and goes into suspend mode.

All hypothesis, speculation , etc.
I'll get the latest version and play around with it some more on the weekend.
- ed/

[AMDave]

Another observation:
Lots and lots of HDD I/O
not related to swapfile
loads of RAM free
appears to be small overhead
it is an odd beastie.

[AMDave]

Some I/O problems on their server:

Thu 20 May 2010 20:43:27 EST orbit@home Finished upload of LINEAR_2009321_10302_1274314780_0_0
Thu 20 May 2010 20:48:06 EST orbit@home Computation for task LINEAR_2009321_10757_1274314785_1 finished
Thu 20 May 2010 20:48:08 EST orbit@home Started upload of LINEAR_2009321_10757_1274314785_1_0
Thu 20 May 2010 20:48:29 EST orbit@home Finished upload of LINEAR_2009321_10757_1274314785_1_0
Thu 20 May 2010 20:52:12 EST Already attached - deleting project_init.xml
Thu 20 May 2010 20:52:24 EST orbit@home update requested by user
Thu 20 May 2010 20:52:28 EST orbit@home Sending scheduler request: Requested by user.
Thu 20 May 2010 20:52:28 EST orbit@home Reporting 2 completed tasks, not requesting new tasks
Thu 20 May 2010 20:53:13 EST orbit@home Scheduler request failed: HTTP internal server error
Thu 20 May 2010 20:53:43 EST orbit@home update requested by user
Thu 20 May 2010 20:53:43 EST orbit@home Sending scheduler request: Requested by user.
Thu 20 May 2010 20:53:43 EST orbit@home Reporting 2 completed tasks, not requesting new tasks
Thu 20 May 2010 20:54:08 EST orbit@home Scheduler request completed
Thu 20 May 2010 20:54:08 EST orbit@home Message from server: Completed result LINEAR_2009321_10302_1274314780_0 refused: result already reported as success
Thu 20 May 2010 20:54:08 EST orbit@home Message from server: Completed result LINEAR_2009321_10757_1274314785_1 refused: result already reported as success

but the results did save and are pending validation so otherwise it all looks good.

[NeoGen]

I just got my first credits on this project, 17.58 for 1 Workunit that took about 110 minutes!