climateprediction.net home page
Detaching
Detaching
log in

Advanced search

Questions and Answers : Windows : Detaching

Author Message
Katherine
Send message
Joined: 12 May 13
Posts: 1
Credit: 241,366
RAC: 81
Message 52807 - Posted: 7 Nov 2015, 18:59:00 UTC

Detaching. I have no option to detach, remove is greyed out and right clicking on the projects yields nothing. After review of the FAQ I have come to the conclusion that this project wants to exclude windows users, there are over 11k of projects for Linux platforms and exactly 0 for windows. It has been like this for some time and I'm sure there are other boinc projects that will be glad to make use of my unused computing time. I have made use of the options to stop getting work but I'd like to get it off my computer. Anyone else know how to get this done?

Les Bayliss
Volunteer moderator
Send message
Joined: 5 Sep 04
Posts: 6408
Credit: 16,737,553
RAC: 27,166
Message 52808 - Posted: 7 Nov 2015, 19:57:43 UTC - in response to Message 52807.

If the option is greyed out, then you attached by means of a different Manager, perhaps BAM! If so then you MUST use the same manager to disconnect.

*****************

The project has been down for nearly 3 weeks, due to a major infrastructure failure in the cabinet housing our servers. One week for the company to rebuild/replace the hardware and for the IT people to bring the software back on line, 3-4 days of people uploading their backlog of files, and a week of firmware updates, and to bring the software back up again.

During this time no new data sets were even considered for loading onto the servers. Hopefully the last few problems will be fixed early in this coming week.

New models for new areas have been under test in the beta site recently, and when some bugs have been fixed, this should result in more work, mostly for Windows.

This post should be read in relation to what model types get work on a given OS.

This project has 3-5 times as many computers wanting work as there are data sets to run. And it most likely will always be this way.


Profile JIM
Send message
Joined: 31 Dec 07
Posts: 982
Credit: 14,189,103
RAC: 18,950
Message 52810 - Posted: 8 Nov 2015, 1:24:33 UTC - in response to Message 52807.

jrapdx
Send message
Joined: 4 Jul 15
Posts: 63
Credit: 2,328,240
RAC: 1,619
Message 52812 - Posted: 8 Nov 2015, 9:24:38 UTC - in response to Message 52810.

> Problem with incapabilities between 64 bit versions of Linux and the 32 bit nature of CPDN have caused many crashes and the need to reissue the same tasks over and over again.

Seems an old story re: 32bit program failures on 64bit Linux machines. It's hard to believe Linux users would be unaware of the 32b on 64b issue. When I decided to run CPDN on Linux, the requirement for having 32bit libs was quite obvious.

Which 32b libs was a bit of a puzzle, but I solved this by installing all of the suggested Ubuntu lib32* packages (and a number of others just to be sure). After that it worked fine.

Actually, until the 32b libs were installed the CPDN models wouldn't even download/start, so I'm not understanding how lack of the libs causes a "compute failure" (unless the request for work itself counts as a failure when 32b libs are missing).

Ultimately the backlog of Linux tasks reflects having too few usable Linux computers in the pool. Not sure what inhibits creating tasks as 64bit binaries, but given the predominance of x86_64 these days, it seems it would be the effective solution.

Profile Dave Jackson
Send message
Joined: 15 May 09
Posts: 1783
Credit: 2,669,678
RAC: 581
Message 52813 - Posted: 8 Nov 2015, 11:53:45 UTC

Actually, until the 32b libs were installed the CPDN models wouldn't even download/start, so I'm not understanding how lack of the libs causes a "compute failure" (unless the request for work itself counts as a failure when 32b libs are missing


Unfortunately the system isn't clever enough. Lots of people have missing libs and still download work which then fails within about 6 seconds. Many of these people have adopted a, "set and forget" attitude to using BOINC and don't notice. Recently a user with 64 cores was identified as downloading a work unit a day for each core which then promptly failed. When identified, work is stopped for these boxes and an email sent out to the address associated with the account but it is taking a long time and I suspect more non working linux boxes may be getting added than are being identified!.

Profile JIM
Send message
Joined: 31 Dec 07
Posts: 982
Credit: 14,189,103
RAC: 18,950
Message 52814 - Posted: 8 Nov 2015, 18:05:10 UTC - in response to Message 52813.

Profile Dave Jackson
Send message
Joined: 15 May 09
Posts: 1783
Credit: 2,669,678
RAC: 581
Message 52815 - Posted: 8 Nov 2015, 21:57:48 UTC - in response to Message 52814.

Profile JIM
Send message
Joined: 31 Dec 07
Posts: 982
Credit: 14,189,103
RAC: 18,950
Message 52818 - Posted: 9 Nov 2015, 5:02:41 UTC - in response to Message 52815.

Profile tullio
Send message
Joined: 6 Aug 04
Posts: 233
Credit: 520,623
RAC: 761
Message 52823 - Posted: 9 Nov 2015, 13:30:23 UTC
Last modified: 9 Nov 2015, 13:31:33 UTC

My Windows 10 is being updated every month, so I have to suspend all BOINC work on it. I have also to restart manually the CERN 2015 Challenge running on Virtual Box after every reboot.
Tullio
____________

Questions and Answers : Windows : Detaching


Main page · Your account · Message boards


Copyright © 2017 climateprediction.net