What is it with this Project, no natter how low I set my Resource Share for this Project it continually dumps more & more Wu's on me.
I have other Projects to run too & don't want or need all these Wu's, Abort City is all I can say.
Right now I have my Resource Share set to 1 for this Project and as high as 7000 at other projects but I can guarantee you sometime during the night I will get dumped on again ... 0_o
Each time you abort a task it improves the chances that the next task you get will be from the same project. Aborts do not adjust LTD like a completed task would. Let your client complete the tasks it downloads and you should not see another task for a long time. A long queue will also cause more tasks to be downloaded than you might expect.
Check your clients for overriding preferences. Global_prefs_override.xml should not have any effect on your resource shares but might anyway. You could have also set overriding project preferences at BAM! or GridRepublic.
____________
BOINC WIKI
BOINCing since 2002/12/8
The whole problem is even though I have my Resource Share set to 1 for this Project and my Connect time to 0.1 if I run out of work from the other Projects then Seti Beta will dump 1 days worth of work on me.
50-60 Wu's simply do not equate to a 1 Resource Share & a 0.1 Connect time in my book anyway. Then when theres work available from the other Projects they won't give me any because I have all these Seti Beta Wu's sitting on my hard drives.
Then by the time I run out the Seti Beta Wu's the other Projects don't have any work again, so what happens, Seti Beta decides to give me anther days worth of work.
It's like a vicious circle as far as I'm concerned, Seti Beta has you Attached & it's going to make darn sure you run their Wu's only ... hehehe
I was correct in assuming I would get dumped on again over night, some of my PC's heavy and some to a lesser degree, so now I'm stuck with running them or Aborting them again so I can get work from the other Projects.
It doesn't matter if I run them either, for every 1 I turn in the Project gives me 1 or 2 or even more right away as soon as I turn it in. I can't win for loosing with this Project.
Then on the other end of the spectrum I'm Attached to the Cosmology Project, now there I can't get Wu's, my PC's keep calling for work but the Project refuses to give half of them any work no matter what I set my Preferences to.
I can create a separate directory with the Cosmology Project being the only 1 Attached to and the Project will just let the PC sit Idle rather than give it any work even though the Server says it has Thousands of Wu's to send out, and if I check other peoples PC's they have hundreds of them waiting to run on theirs...Go figure
I never had these problems when I was running v5.4.11, but as soon as I started Updating to the v5.8 & above it's been nothing but a nightmare trying to either get work or to keep a project from giving me to much ...
You might check your general preferences there is a new setting that applies to the newer clients.
Maintain enough work for an additional x.xx days
(Requires 5.10+ client.)
____________
The whole problem is even though I have my Resource Share set to 1 for this Project and my Connect time to 0.1 if I run out of work from the other Projects then Seti Beta will dump 1 days worth of work on me.
...
I never had these problems when I was running v5.4.11, but as soon as I started Updating to the v5.8 & above it's been nothing but a nightmare trying to either get work or to keep a project from giving me to much ...
The project merely fulfills whatever amount of work BOINC requests. It is likely the long term debts for the projects are badly imbalanced, you might want to get BoincDV to check that, and possibly to reset them. LTD is really long term and must be to handle projects like CPDN with work that lasts months or years, so seldom does what is expected over spans of days or even weeks.
You might check your general preferences there is a new setting that applies to the newer clients.
Maintain enough work for an additional x.xx days
(Requires 5.10+ client.)
Yes, I know about that Setting as I've seen it at the projects & now with the v5.10 & above installed across my PC's in the BOINC Manager Preferences.
As far as I'm concerned if it wasn't confusing enough before on how much work your asking for adding that setting makes it even more confusing. I'm trying to let my Local Preference Settings take Precedence over the Projects Settings using the Preference Tab I guess you'd call it in the BOINC Manager but I'm not sure if it's even taking effect or not ...
All I know is when I woke up this morning several of my PC's where just spinning their wheels because they had no work to do even though they were calling for work from the Projects. So right now it's ether I get a Boatload of Seti Beta WU's or I get nothing at all, if the PC runs out then it just sits there doing nothing even though their all Attached to 5 or 6 Projects and calling for work from all of them ... :(
The whole problem is even though I have my Resource Share set to 1 for this Project and my Connect time to 0.1 if I run out of work from the other Projects then Seti Beta will dump 1 days worth of work on me.
...
I never had these problems when I was running v5.4.11, but as soon as I started Updating to the v5.8 & above it's been nothing but a nightmare trying to either get work or to keep a project from giving me to much ...
The project merely fulfills whatever amount of work BOINC requests. It is likely the long term debts for the projects are badly imbalanced, you might want to get BoincDV to check that, and possibly to reset them. LTD is really long term and must be to handle projects like CPDN with work that lasts months or years, so seldom does what is expected over spans of days or even weeks.Joe
I have ran the BoincDV for a long time with no change in getting work, it doesn't seem to matter if the Debt's are even or not, the Projects that won't give me work still won't give me work. The Manager is calling for work from them but for whatever reason the Project refuses to send it, no reason given in the Manager either, just the No Work From Project Message when I know theres work available ...
EXAMPLE: Cosmology has 3,700 WU's available right now but this is what I get when calling for work to the Project, this PC hasn't seen a Cosmo WU in day's when some of my other PC's have downloaded many WU's from there ...
Computer Host ID Project Date Message
PBT-9 MGR1 32-Bit PBOYZTOY9 386 Cosmology@Home 6/21/2007 9:33:30 AM Reason: no work from project
PBT-9 MGR1 32-Bit PBOYZTOY9 385 Cosmology@Home 6/21/2007 9:33:30 AM Deferring communication for 2 hr 55 min 25 sec
PBT-9 MGR1 32-Bit PBOYZTOY9 384 Cosmology@Home 6/21/2007 9:33:30 AM Reason: requested by project
PBT-9 MGR1 32-Bit PBOYZTOY9 383 Cosmology@Home 6/21/2007 9:33:30 AM Deferring communication for 7 sec
PBT-9 MGR1 32-Bit PBOYZTOY9 382 Cosmology@Home 6/21/2007 9:33:30 AM Message from server: No work sent
PBT-9 MGR1 32-Bit PBOYZTOY9 381 Cosmology@Home 6/21/2007 9:33:30 AM Scheduler RPC succeeded [server version 510]
PBT-9 MGR1 32-Bit PBOYZTOY9 380 Cosmology@Home 6/21/2007 9:33:25 AM Requesting 71577 seconds of new work
PBT-9 MGR1 32-Bit PBOYZTOY9 379 Cosmology@Home 6/21/2007 9:33:25 AM Sending scheduler request: Requested by user
What is it with this Project, no natter how low I set my Resource Share for this Project it continually dumps more & more Wu's on me.
I have other Projects to run too & don't want or need all these Wu's, Abort City is all I can say.
Right now I have my Resource Share set to 1 for this Project and as high as 7000 at other projects but I can guarantee you sometime during the night I will get dumped on again ... 0_o
one easy way around this I've found is to go into Projects and select no new tasks it's worked ok over the last month even with spinhenge who have a habit of sending 20 wus. Being a non technical type I find the easy way is sometimes the best. When I get down again to a resonable level just do the reverse go into projects and allow new work.
____________
A computer program will always do what you tell it to do, but rarely what you want it to do.
one easy way around this I've found is to go into Projects and select no new tasks it's worked ok over the last month even with spinhenge who have a habit of sending 20 wus. Being a non technical type I find the easy way is sometimes the best. When I get down again to a resonable level just do the reverse go into projects and allow new work.
Problem with that is then I end up with what I had this morning, no work at all on a few PC's. They were calling for work but none of the other Projects I'm Attached to would give them any even though there was work available & Seti Beta wasn't calling for work so they went dry.
I'm seriously thinking about going back to v5.8.16, ever since I installed v5.10.0 & v5.10.7 yesterday on a my PC's 3 of them have crashed with the BSOD, 1 of them a few times already. I never had that happen with v5.8.16.
Their not showing any errors or anything, just sitting there Idle with a Blue Screen Showing ...
one easy way around this I've found is to go into Projects and select no new tasks it's worked ok over the last month even with spinhenge who have a habit of sending 20 wus. Being a non technical type I find the easy way is sometimes the best. When I get down again to a resonable level just do the reverse go into projects and allow new work.
Problem with that is then I end up with what I had this morning, no work at all on a few PC's. They were calling for work but none of the other Projects I'm Attached to would give them any even though there was work available & Seti Beta wasn't calling for work so they went dry.
I'm seriously thinking about going back to v5.8.16, ever since I installed v5.10.0 & v5.10.7 yesterday on a my PC's 3 of them have crashed with the BSOD, 1 of them a few times already. I never had that happen with v5.8.16.
Their not showing any errors or anything, just sitting there Idle with a Blue Screen Showing ...
There is little difference, besides minor tweaks, in the way V5.8 and V5.10 does work fetch other that then 5.10 allow you to top off your cache with extra XX days work. If your cache size is kept at a resonable .5 plus 2 days extra you should have no problems. From what I read you're micromanaging Boinc. Boinc works as expected if you leave it alone. Changing parameters every day or two or three will only mess up the scheduler. It take time to adjest to the changes.
Just because a projects says it has work when you look at it doesn't mean that there is work on hand when Boinc calls in. Few project have real time updates of this information. Set your preferences to what you want and LEAVE BOINC ALONE for a week or two and things will settle down.
The Scheduler is very good at maintaining resource share and work queue. Most of the projects run out of work from time to time but Boinc compensates for that later. I've never run out of work on the system I run off line and reconnect every day or two. The project I set to a very low resource only get work every couple of months as intended, while the others continue as expected. I can only presume that your continual tinkering with the setting is the problem.
____________
Questions? Answers are in the BOINC Wiki.
Boinc V6.10.45 Test
WinXP C2D 2.1G 3GB
I don't feel I'm Micro Managing anything, I'm asking the BOINC Manager to do what I want it to do and not have a mind of it's own and to hell with what I want. If I want to change the Preferences to favor 1 Project over another then that's what BOINC should do, & it should do it now, not in a week or 2. It should not feel that I owe some other Project which I don't want to run anymore more or run less and force me to continue to run that Project.
This is exactly what is driving a lot of people back to or over to the DC Projects, at least when you make a change at a lot of those projects it's instantly recognized and implemented. I feel that the Debt Sharing thing with the BOINC Projects is 1 of the worst things ever implemented, I've felt that for a long time, it simply screw's every thing up when you want to run some other Project.
I shouldn't have to wait 2 or 3 days or even more for BOINC to decide it's okay to run that Project now, by the time BOINC makes up it's mind to let you run that Project their probably out of work anyway so your left out then. I guess your saying the next time which I figure won't be too long that 1 of my PC's which is Attatched to 6 Diff Projects runs out of work to just let it sit there and don't mess with anything, BOINC knows best & will decide when it's time to give it more work ... 0_o
And as a side note BOINC v5.10.0 has crashed yet another PC of mine, so thats the 4'th PC it has managed to put down, a great BOINC Release is all I can say.
6/21/2007 9:33:25 AM Requesting 71577 seconds of new work
PBT-9 MGR1 32-Bit PBOYZTOY9 379 Cosmology@Home 6/21/2007 9:33:25 AM Sending scheduler request: Requested by user
Hmm, why are you asking for close to 20 hours of work if you're running with a 0.1 days cache-setting as you claim to be doing, and AFAIK you've not got a 9-way or more-cpu-system?
As for not getting work from Cosmology@home, it looks like you've already got over 10% of "in progress", so my guess is you've already got another copy from the same wu's and therefore there's no available work you can get...
6/21/2007 9:33:25 AM Requesting 71577 seconds of new work
PBT-9 MGR1 32-Bit PBOYZTOY9 379 Cosmology@Home 6/21/2007 9:33:25 AM Sending scheduler request: Requested by user
Hmm, why are you asking for close to 20 hours of work if you're running with a 0.1 days cache-setting as you claim to be doing, and AFAIK you've not got a 9-way or more-cpu-system?
As for not getting work from Cosmology@home, it looks like you've already got over 10% of "in progress", so my guess is you've already got another copy from the same wu's and therefore there's no available work you can get...
Didn't know I was asking for 20 hr's of work, but I do keep my settings @ 0.1 to 0.3 connect Time. But being that the Computer is a Quad Core if my figuring is right thats only about 5 hr's per Core, correct me if I'm wrong. I've also gone as low as 0.0001 Connect Time with the same results, no work from Cosmology but a cache of 30 to 50 from Seti Beta.
From what I've noticed with Seti Beta by using BOINCView is if theres no other work from the other Projects in the Manager it shows Seti Beta as 100% then, and thats when it will just fill up Manager with Wu's.
Running Seti Beta I don't dare go over 0.1 to 0.3 Connect Time or I would have 10 days of Seti Beta Downloaded to me probable.
I don't know what your talking about with the 10% of "in progress" jargon, please explain a little more, maybe thats where my problem lies & I don't have a clue either about another copy ... ???
I don't know what your talking about with the 10% of "in progress" jargon, please explain a little more, maybe thats where my problem lies & I don't have a clue either about another copy ... ???
Most projects sends-out multiple copies of same wu, Cosmology@home seems to send-out 4 copies/wu. To guard against invalid results, the same user is normally not allowed to download more than one of these copies from the same wu.
Meaning, if example your computer-#1 downloads result-0 of wu-1234, your computer-#2 is now allowed to download result-1, 2 or 3 of wu-1234.
With your computers having 250+ results "in progress", this means potentially 750+ results in "ready to send"-queue you can't download. Due to little available work and few users, this is a large fraction of the total.
____________
Okay, I understand now, thanks for the explanation, that may be the reason at times I can't get any work from there ...
I'm seriously thinking about going back to v5.8.16,
I always wait a long time before upgrading so I can see all the problems others have first. I finally upgraded to 5.8.16 from 5.4.11 because cosmo and lattice required it.
____________
