WU Started then switch to new computer no results posted

Message boards : Number crunching : WU Started then switch to new computer no results posted

To post messages, you must log in.

AuthorMessage
halfmeg

Send message
Joined: 14 Dec 05
Posts: 7
Credit: 2,496
RAC: 0
Message 9227 - Posted: 17 Jan 2006, 19:59:15 UTC
Last modified: 17 Jan 2006, 20:00:38 UTC

Hi folks, I have 3 computers here. One is kind of old 500Mhz, the other two both over 2Ghz.

I allowed work on the slow system and a WU

( https://boinc.bakerlab.org/rosetta/result.php?resultid=7035493 )

arrived which was somewhat different than the normal WU I have seen. It incremented in 2.5% chunks instead of 10% and would have run over 40 hours if I had left it in place. I suspended it, thinking I wouldn't get any more work, but another WU downloaded

( https://boinc.bakerlab.org/rosetta/result.php?resultid=7145133 ).

I suspended the project and then the 2nd WU. I then transfered the entire BOINC directory to one of the faster computer and allowed them to process on it overnight. They both completed successfully. The 1st one is still odd, in excess of 30 red result dots in the graph before it finished.

The 2nd computer uploaded the results but nothing on my results page changed ( shows outcome unknow still ). Thinking perhaps the computer that downloaded the WU has to report them I had already transfered the entire BIONIC directory back to the slow computer before the upload. It also successfully reported them back to the server, but still no change in my results page.

Although out of the ordinary, should this have worked ( downloading to one, processing on another, and uploading by either )?

I don't care much about the credit, but don't like the idea that orphaned WUs ( status unknown to servers but known to be lost, destroyed, botched, ______ ) have to wait almost a month before being rescheduled for processing. Can a reset be placed on our results page to free the orphaned WUs back into the pool?

Oh yeah, what kind of WU has 30+ end results or whatever you call them?

Phil
ID: 9227 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Divide Overflow

Send message
Joined: 17 Sep 05
Posts: 82
Credit: 921,382
RAC: 0
Message 9240 - Posted: 18 Jan 2006, 2:27:00 UTC - in response to Message 9227.  
Last modified: 18 Jan 2006, 2:44:33 UTC


Although out of the ordinary, should this have worked ( downloading to one, processing on another, and uploading by either )?

No.

You should have just let it crunch on the original machine. (If you didn't want any Rosetta work for that host, the command that you were looking for was the suspend button from the projects tab.) Don't worry about that WU any longer. It will eventually expire at it's deadline and be sent out to another host. BOINC is not setup for transferring assigned and downloaded WU's from one host to another.

The new WU's appear to be checkpointing more often, which means smaller % complete increases and lots more of those "endpoints".

ID: 9240 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Snake Doctor
Avatar

Send message
Joined: 17 Sep 05
Posts: 182
Credit: 6,401,938
RAC: 0
Message 9310 - Posted: 18 Jan 2006, 22:15:24 UTC - in response to Message 9227.  

Hi folks, I have 3 computers here. One is kind of old 500Mhz, the other two both over 2Ghz.

I allowed work on the slow system and a WU

( https://boinc.bakerlab.org/rosetta/result.php?resultid=7035493 )

arrived which was somewhat different than the normal WU I have seen. It incremented in 2.5% chunks instead of 10% and would have run over 40 hours if I had left it in place. I suspended it, thinking I wouldn't get any more work, but another WU downloaded

( https://boinc.bakerlab.org/rosetta/result.php?resultid=7145133 ).

I suspended the project and then the 2nd WU. I then transfered the entire BOINC directory to one of the faster computer and allowed them to process on it overnight. They both completed successfully. The 1st one is still odd, in excess of 30 red result dots in the graph before it finished.

The 2nd computer uploaded the results but nothing on my results page changed ( shows outcome unknow still ). Thinking perhaps the computer that downloaded the WU has to report them I had already transfered the entire BIONIC directory back to the slow computer before the upload. It also successfully reported them back to the server, but still no change in my results page.

Although out of the ordinary, should this have worked ( downloading to one, processing on another, and uploading by either )?

I don't care much about the credit, but don't like the idea that orphaned WUs ( status unknown to servers but known to be lost, destroyed, botched, ______ ) have to wait almost a month before being rescheduled for processing. Can a reset be placed on our results page to free the orphaned WUs back into the pool?

Oh yeah, what kind of WU has 30+ end results or whatever you call them?

Phil


Phil.

If you look at your stats page, you will see that the stats are tied to the particular computer. That is why what you attempted did not work. the computer that downloads the WU is expected to return the result. Unless this occurs, the WU will be listed as unreported when the deadline arrives. At that point it will be sent to another machine for processing. If you look at your account page and click on "view computers" you will see that separate stats are kept for each machine. That is just how BOINC works.

Regards
Phil (also)


We Must look for intelligent life on other planets as,
it is becoming increasingly apparent we will not find any on our own.
ID: 9310 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
keputnam

Send message
Joined: 18 Sep 05
Posts: 24
Credit: 2,084,465
RAC: 0
Message 9320 - Posted: 19 Jan 2006, 4:11:03 UTC - in response to Message 9227.  

What you could have done was transfer the directory as you did, but disable network access in BOINC on the fater computer, then when the work units finished, copy the directory back to the slow machine and let it report from there.

A WU must be returned by the machine that downloaded it to get credit for it


ID: 9320 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile River~~
Avatar

Send message
Joined: 15 Dec 05
Posts: 761
Credit: 285,578
RAC: 0
Message 9332 - Posted: 19 Jan 2006, 10:45:18 UTC - in response to Message 9320.  

What you could have done was transfer the directory as you did, but disable network access in BOINC on the fater computer, then when the work units finished, copy the directory back to the slow machine and let it report from there.

This should not be done except in an emergency (eg the first machine dies and the hard disk is salvaged).

Some projects don't mind mixed machines crunching a work unit, some do. In particular LHC and Predictor explicitly say they don't want this sort of mixed processing - Predictor for example makes sure that all results in a given WU go to the same kind of machine, and you defeat that intention by this work around.

In general run the code as it is designed to be run, then the project and the programmers know what is going on. It would, for example, confuse a database scan intended to spot bugs specific to (say) HT if you ran on an HT machine but seemed to run on a non-HT machine.


A WU must be returned by the machine that downloaded it to get credit for it

er no. If you copied the directory across and left it on the faster machine, in some cases and with some projects it will work. But in view of what Ive said above, this again is only an emergency measure and only when are sure the project would like that. CPDN, with their very long WU, prefer you do this rather than lose weeks of work - most other projects as far as I know prefer you not to.

River~~

ID: 9332 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile carl.h
Avatar

Send message
Joined: 28 Dec 05
Posts: 555
Credit: 183,449
RAC: 0
Message 9343 - Posted: 19 Jan 2006, 12:34:36 UTC

From Bill Michaels FAQ

Q. Can I move work from one computer to another?
A. Yes. If you copy the ENTIRE Boinc folder to a thumb drive, take that to a different system, plug it in and run, it will continue to work. The only downside to doing this is that your "view computers" display may get a little confused if you upload a WU from the new system. This can be prevented by running BOINC Manager from the thumbdrive initially after the copy, and setting network access to never available.

Not all Czech`s bounce but I`d like to try with Barbar ;-)

Make no mistake This IS the TEDDIES TEAM.
ID: 9343 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : WU Started then switch to new computer no results posted



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