| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Computing for Sustainable Water Forum
Thread: Known Issue with Linux stuck workunits [Resolved] |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 49
|
|
| Author |
|
|
TKH
Former World Community Grid Admin USA Joined: Aug 13, 2005 Post Count: 775 Status: Offline Project Badges: 
|
Hello Moonian,
Will try and replace the images for you later today. Thanks! Tedi |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
3. To create a mechanism to allow for linux users to choose how many they want for cfsw would take time and divert energy away from actually solving the issue at hand. Thanks, -Uplinger I think with the memory requirements of this project both windows and Linux this may be very useful like it was for Clean Energy 2. I have a Windows 7 machine running 6 of these and it is making it difficult to use that machine. I would like to see something like this implemented as it may be hard to keep the right balance of these running even with a mix of projects. It also seems a windows xp machine running 12 cores(dual 6 core opterons) and to many cores starting this project at once may be causing work units to error. Thanks for the help and info on the stuck work units. I'll be interested to see the fix for this or if its an issue with the 64 bit Ubuntu. Thanks again, Jason My Octo threaded laptop runs 8 concurrent at 100% CPU timer, with a workset of 279MB per task without any hassle whatsoever. I'd expect that a 6 core device would have at least 3GB usable RAM, which is more than sufficient. If you have trouble running 6 concurrent, then open a new thread with your settings [post first 35 lines of message log after boot], and then we'll do some 'constructive' criticizing on what may be improved. One of the ways to control how many big jobs run concurrent is limiting BOINC memory use [options are for during use / during idle]. That way the smaller will run always, and if too many biggies are used, one or more of these are put in "waiting for memory", to let the smaller ahead. --//-- edit: We're really going off topic here [this needs a new thread!], but this Process Explorer properties screenshot shows the said memory use, here 277MB peak after 37% progress, and a virtual memory use of flat 300MB. The Page fault delta is negligeble, suggesting that the VM access access is next to non-existent, which may be where your problem is sitting... too much disk swapping due to whatever constraint. Getting average efficiency of 99.85% when system is left idle and 98.5% when system is used i.e. user impairment not perceivable.  [Edit 2 times, last edit by Former Member at Apr 24, 2012 5:57:38 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello SekeRob,
----------------------------------------Reference: SekeRob [Apr 24, 2012 1:36:18 PM] post Are you referring to CEP2 or CFSW in your "... 8 concurrent at 100% CPU timer, with a workset of 279MB per task without any hassle whatsoever."? ------------------- edit1: Is your edit done Apr 24, 2012 5:57:38 PM an attempt to answer the above question in that you are referring to CFSW in your "... 8 concurrent at 100% CPU timer, with a workset of 279MB per task without any hassle whatsoever."? ; ; edit1_2012.04.24Tu.1857.utc ; [Edit 1 times, last edit by Former Member at Apr 24, 2012 6:55:41 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello Moonian, Will try and replace the images for you later today. Thanks! Tedi I see you've helped me edited the post. Thanks for your help  |
||
|
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges: 
|
!
----------------------------------------[Edit 1 times, last edit by skgiven at Jul 18, 2012 9:09:52 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello skgiven
----------------------------------------Reference: skgiven [Apr 27, 2012 8:22:24 PM] post From a blindly-crunching cruncher, all of what you have discussed does not matter. No emergencyProfile? No problem, and who cares anyway? No projectStatus? A projectStatus being an approach to reduce the need for an emergencyProfile? What was that again? Ignorance is bliss. For those crunchers at the other side of the spectrum, the lack of a projectStatus webpage and/or a serverStatus webpage makes crunching difficult. Ignorance is painful. ; ; edit1_2012.04.28Sa.1354.utc > readability improvements ; [Edit 1 times, last edit by Former Member at Apr 28, 2012 1:54:04 PM] |
||
|
|
Jason1478963
Senior Cruncher United States Joined: Sep 18, 2005 Post Count: 295 Status: Offline Project Badges: 
|
3. To create a mechanism to allow for linux users to choose how many they want for cfsw would take time and divert energy away from actually solving the issue at hand. Thanks, -Uplinger I think with the memory requirements of this project both windows and Linux this may be very useful like it was for Clean Energy 2. I have a Windows 7 machine running 6 of these and it is making it difficult to use that machine. I would like to see something like this implemented as it may be hard to keep the right balance of these running even with a mix of projects. It also seems a windows xp machine running 12 cores(dual 6 core opterons) and to many cores starting this project at once may be causing work units to error. Thanks for the help and info on the stuck work units. I'll be interested to see the fix for this or if its an issue with the 64 bit Ubuntu. Thanks again, Jason My Octo threaded laptop runs 8 concurrent at 100% CPU timer, with a workset of 279MB per task without any hassle whatsoever. I'd expect that a 6 core device would have at least 3GB usable RAM, which is more than sufficient. If you have trouble running 6 concurrent, then open a new thread with your settings [post first 35 lines of message log after boot], and then we'll do some 'constructive' criticizing on what may be improved. One of the ways to control how many big jobs run concurrent is limiting BOINC memory use [options are for during use / during idle]. That way the smaller will run always, and if too many biggies are used, one or more of these are put in "waiting for memory", to let the smaller ahead. --//-- edit: We're really going off topic here [this needs a new thread!], but this Process Explorer properties screenshot shows the said memory use, here 277MB peak after 37% progress, and a virtual memory use of flat 300MB. The Page fault delta is negligeble, suggesting that the VM access access is next to non-existent, which may be where your problem is sitting... too much disk swapping due to whatever constraint. Getting average efficiency of 99.85% when system is left idle and 98.5% when system is used i.e. user impairment not perceivable. Thanks for the input on this. I currently have 2gb of ram in that machine and have zero plans to buy more DDR2 memory as it does everything fine without this project running. If there were more then four profiles I may be able to give that a try. I have older systems that just crunch and require the percentage of memory to be set high to utilize all cores. It is easier for me and possibly many others if the techs follow what they did with Clean Energy 2 or to leave the project for others to run. It seems you lose to much work if you don't leave applications in memory with projects like Clean Energy 2. This situation seems to make the in use memory setting and idle memory setting less helpful. If you run Linux on most crunchers 2gb is overkill most of the time on even 6 and 8 cores. The 3gb limit on memory with windows XP may have pushed me into the 2gb quad core builds that were later updated to 6 cores.  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Any word on a fix for the stuck WU's yet? I would love to offer more core resources to this project than just one WU per machine at a time. I understand that the techs have been busy with upgrades but I only have Linux devices available so I am starving
 |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks for the input on this. I currently have 2gb of ram in that machine and have zero plans to buy more DDR2 memory as it does everything fine without this project running. If there were more then four profiles I may be able to give that a try. I have older systems that just crunch and require the percentage of memory to be set high to utilize all cores. It is easier for me and possibly many others if the techs follow what they did with Clean Energy 2 or to leave the project for others to run. It seems you lose to much work if you don't leave applications in memory with projects like Clean Energy 2. This situation seems to make the in use memory setting and idle memory setting less helpful. If you run Linux on most crunchers 2gb is overkill most of the time on even 6 and 8 cores. The 3gb limit on memory with windows XP may have pushed me into the 2gb quad core builds that were later updated to 6 cores. Efforts to program a selection option for X number of tasks of any science are unlikely. CEP2 is the exception because it is causing for the unsuspected and on many devices IO issues. It's a driver behind the thought path of studying suitability of multi-threaded processing, so that e.g. a quad core would occupy all threads for 1 task [and run X times faster], and by that token reduce the overall memory footprint and the IO loads of course. Till then, think those that consciously went for 2 GB on a quad are few and far between, let alone 6 core upgraders. [And hope those also consciously went for 2 x 1GB strips instead of 1 x 2GB, for that is giving better performance]. Guess > 95% or more expanded memory, cost small, to improve their Windows performance and not for BOINCing purposes. --//-- http://bit.ly/WCGMTX [Edit 1 times, last edit by Former Member at May 2, 2012 4:58:53 AM] |
||
|
|
armstrdj
Former World Community Grid Tech Joined: Oct 21, 2004 Post Count: 695 Status: Offline Project Badges:
|
We are currently testing a fix in our Alpha testing environment. If things go well, so far so good, I would expect a Beta in the next couple of days. Thanks for everyone's patience while we work through this issue.
Thanks, armstrdj |
||
|
|
|