i dont know for sure... but u can just not let the comp work on it for a bit, the ligo ppl dont want the program to take up too much info on ur comp, so they have a work limit... dont worry about it, enless ur a few hours from the deadline :(
I sometimes get the message "Suspending work fetch because computer is overcommitted" What does that mean?
A lot of work has been done recently by John McLeod VII (JM7) on making sure, automagically, that you don't miss deadlines with your work when you work on multiple projects. There is a new scheduling routine in the latest Core Clients which goes to a lot of trouble to prevent you wasting your cpu cycles by missing deadlines. If this local intelligence senses that you may have too much work (ie be overcommitted) it will temporarily disable fetching new work until it reckons that the danger is over. It also goes to a lot of trouble to honour the way you have allocated your resource share between projects. If it has to give extra time to meet a deadline in one project, it keeps track of this and when the danger has passed it will give extra time to those projects that were not in trouble. Preventing work fetch is part of that process of "balancing the debt".
Whilst JM7 has done a very commendable job of having a system in place that is as close as possible to "idiot proof", if you think of all the possible permutations and combinations of variables that this routine has to cope with, it is little wonder that it needs time to stabilise and work well. The dumbest thing that people do is try to force things by doing multiple manual overrides of what the automatic process is trying to achieve. Then they winge about how bad the scheduling routine is. It's also pretty dumb to have settings that conflict with each other like giving E@H a low resource share and a large "connect" interval.
In a nutshell, if you want it to stabilise quickly, don't have too many projects running on slow machines, choose your resource share sensibly and keep your "connect to network" interval as small as possible. If you were running three to five projects on a reasonably fast machine you might want to play around with values between 0.1 days and perhaps 1.0 days absolute max. If you set a value of 2 to 3 days or more, you'll probably continually find yourself in deadline trouble. The type of network connection, broadband vs dialup also has a big bearing on what you choose for this value.
If you want to know a lot more about this do a search for "Long Term Debt" or "LTD" over at Seti. There's zillions of threads there about how it all works.
Quote:
Should I do anything? Hugh
If you are getting this message repeatedly over a time interval of 7-14 days, your settings are not very BOINC friendly. You probably need to reduce your "connect" interval and/or increase the resource share of the project having deadline trouble. You do however need to let it sort itself out for possibly up to a week or more. The longer it takes to settle, the more "challenging" are the settings you have chosen. You will know when you have got it right. It will settle quickly.
Give more details if you want specific advice. Good luck!!
come here if u cant find the thread w/ the topic u want
)
ask a question, make a statment, whatever, anything goes
Where to go, if I cannot find the topic with the thread I want?
post it here
)
post it here
Babe of the Day™
)
Babe of the Day™
me-[at]-rescam.org
ask a question, make a
)
ask a question, make a statment, whatever, anything goes
Where to go, if I cannot find the topic with the thread I want?
I sometimes get the message "Suspending work fetch because computer is overcommitted" What does that mean? Should I do anything? Hugh
i dont know for sure... but u
)
i dont know for sure... but u can just not let the comp work on it for a bit, the ligo ppl dont want the program to take up too much info on ur comp, so they have a work limit... dont worry about it, enless ur a few hours from the deadline :(
I want to see a thread
)
I want to see a thread entitled "Message Board Search Function Is Coming" :D
EDIT: Ummm... yeah... I finally found it. There is a search function!
http://einstein.phys.uwm.edu/forum_text_search_form.php
RE: I sometimes get the
)
A lot of work has been done recently by John McLeod VII (JM7) on making sure, automagically, that you don't miss deadlines with your work when you work on multiple projects. There is a new scheduling routine in the latest Core Clients which goes to a lot of trouble to prevent you wasting your cpu cycles by missing deadlines. If this local intelligence senses that you may have too much work (ie be overcommitted) it will temporarily disable fetching new work until it reckons that the danger is over. It also goes to a lot of trouble to honour the way you have allocated your resource share between projects. If it has to give extra time to meet a deadline in one project, it keeps track of this and when the danger has passed it will give extra time to those projects that were not in trouble. Preventing work fetch is part of that process of "balancing the debt".
Whilst JM7 has done a very commendable job of having a system in place that is as close as possible to "idiot proof", if you think of all the possible permutations and combinations of variables that this routine has to cope with, it is little wonder that it needs time to stabilise and work well. The dumbest thing that people do is try to force things by doing multiple manual overrides of what the automatic process is trying to achieve. Then they winge about how bad the scheduling routine is. It's also pretty dumb to have settings that conflict with each other like giving E@H a low resource share and a large "connect" interval.
In a nutshell, if you want it to stabilise quickly, don't have too many projects running on slow machines, choose your resource share sensibly and keep your "connect to network" interval as small as possible. If you were running three to five projects on a reasonably fast machine you might want to play around with values between 0.1 days and perhaps 1.0 days absolute max. If you set a value of 2 to 3 days or more, you'll probably continually find yourself in deadline trouble. The type of network connection, broadband vs dialup also has a big bearing on what you choose for this value.
If you want to know a lot more about this do a search for "Long Term Debt" or "LTD" over at Seti. There's zillions of threads there about how it all works.
If you are getting this message repeatedly over a time interval of 7-14 days, your settings are not very BOINC friendly. You probably need to reduce your "connect" interval and/or increase the resource share of the project having deadline trouble. You do however need to let it sort itself out for possibly up to a week or more. The longer it takes to settle, the more "challenging" are the settings you have chosen. You will know when you have got it right. It will settle quickly.
Give more details if you want specific advice. Good luck!!
Cheers,
Gary.
Who'll win Wimbledon this
)
Who'll win Wimbledon this year???