Scheduling issue?

Ascholten
Ascholten
Joined: 20 Oct 10
Posts: 63
Credit: 2848108
RAC: 0
Topic 195479

Yes it's me again. Go ahead, take your prozac, drink a 5th of vodka and attempt to placate me, I understand.

I am currently running on this machine Boinc with Einstein at 80 percent, world community at 10 percent and milkyway at 10 with get no more tasks.

The problem I am seeing is, Einstein wont load up anymore work besides for maybe one or two tasks after the ones that are working. I have asked for a few days worth. (internet here is sh&&&&y and I want to make sure my machine is working even if it can't talk).

Even after changing my numbers from 1 day to 7 days worth of work and even changing my internet access to 7 days, einstein wont load up. World Community sure will but Estein wont.

To get einstein to finally load me with the tasks I had to actually suspend the other projects currently running, even though they had a low percentile of op time.. (10 percent).

Maybe something needs to be done with looking at the projects .vs. the 'sharing' of overall time each is allotted in determining 'daily limits' as to what they can do when determining if to send more jobs or not.

Maybe something is stupid with my machine... I don't know, but thought Id mention it to see if I can get any constructive replies as to a possible fix.

Thank you
Aaron

If god meant for us not to BOINC he'd have made our #$%^%^ shorter!!

Gundolf Jahn
Gundolf Jahn
Joined: 1 Mar 05
Posts: 1079
Credit: 341280
RAC: 0

Scheduling issue?

That might well be a scheduling problem, but then it's a BOINC problem, not an Einstein problem. (BOINC Message boards)

As for your changing the cache parameters: did you adjust them online (in your account) or locally (BOINC manager Advanced menu)? If online, did you check that your local preferences are cleared (button), since they override the online ones?

Gruß,
Gundolf

Computer sind nicht alles im Leben. (Kleiner Scherz)

Ascholten
Ascholten
Joined: 20 Oct 10
Posts: 63
Credit: 2848108
RAC: 0

Ok I didnt check that they

Ok I didnt check that they over rode my online settings. I was under the impresson that your local settings took precedence over what was online, since one may have multi machines and each would be tailored to suit it best.

Locally I set them to what they are and did another update request etc hoping they'd get me new tasks. when just running einstein this always worked, as I increased the 'buffer' id get more of a stack of tasks. In this situation it did not seem to help at all with the einstein project.

Not complaining just making an observation and wondering if this is a known issue and if there was anything I could do about it, except suspend other projects, get the einstein files then restart them.

Thank you
Aaron

If god meant for us not to BOINC he'd have made our #$%^%^ shorter!!

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5872
Credit: 117820781780
RAC: 34754070

RE: Even after changing my

Quote:
Even after changing my numbers from 1 day to 7 days worth of work and even changing my internet access to 7 days, einstein wont load up.


If you set the 'internet access' to 7 days you are effectively telling BOINC that there could be a 7 day delay in your host's ability to report completed work. This tends to cause BOINC to severely limit your work on hand. If your host is essentially 'always on", your best bet is to set that pref very low or even zero. Just use the 'extra days' setting and don't go beserk with it :-).

If you still can't get lots of work, there must be something else wrong.

Cheers,
Gary.

mikey
mikey
Joined: 22 Jan 05
Posts: 12706
Credit: 1839111349
RAC: 3589

RE: Ok I didnt check that

Quote:

Ok I didnt check that they over rode my online settings. I was under the impresson that your local settings took precedence over what was online, since one may have multi machines and each would be tailored to suit it best.

Locally I set them to what they are and did another update request etc hoping they'd get me new tasks. when just running einstein this always worked, as I increased the 'buffer' id get more of a stack of tasks. In this situation it did not seem to help at all with the einstein project.

Not complaining just making an observation and wondering if this is a known issue and if there was anything I could do about it, except suspend other projects, get the einstein files then restart them.

Thank you Aaron

Okay do this...go into the Boinc Manager, down by the clock, and go to the Projects tab and click on each Project and then on the left click on properties for it. Down at the bottom of that screen that pops up there are some lines I need the numbers from: the cpu scheduling priority and the cpu work fetch priority. I think what is happening is you are running into a Long Term Debt issues and Boinc is doing exactly what it is supposed to based on your settings.
One more thing what is the percentage resource share for each project show in the Projects tab? Based on those answers I think I can tell you what is happening and why Einstein is not downloading any work right now. AND why when you suspend the other projects Einstein DOES get work!

Ascholten
Ascholten
Joined: 20 Oct 10
Posts: 63
Credit: 2848108
RAC: 0

Ok let me see here for

Ok let me see here
for scheduling. percent of resource share.
einstein gets 80 percent
world community gets 10 percent
milkyway gets 10 percent however is currently set to accept no new tasks and is not working on anything.

einstein CPU scheduling 1288
work fetch -37658

world community cpu scheduling -2425
cpu work fetch 0.

milky way currently suspended and set to fetch no more work
cpu scheduling 1144
cpu work fetch 0.

I am not looking to 'go crazy' with the tasks Id just like to have a few days on the machine so that if my internet gets stupid I have a 'buffer' of work to do until it comes back online 12 hours or 2 days later.

I generally keep my internet set to always available, and access connect about every .1 to 1.0 days. currently set to 1.0 days.

I will wait until the work I have flushes out a bit before i start tweaking, I dont want to goof with settings and load up on 10 days worth of work if I can help it.

Thanks again.
Aaron

If god meant for us not to BOINC he'd have made our #$%^%^ shorter!!

Gundolf Jahn
Gundolf Jahn
Joined: 1 Mar 05
Posts: 1079
Credit: 341280
RAC: 0

RE: einstein CPU

Quote:
einstein CPU scheduling 1288
work fetch -37658


That's definitely a Long Term Debt issue.

If you don't agree with BOINC's assessment, you could zero that value for all projects by setting (temporarily) 1 in cc_config.xml.

Quote:
Set resource scheduling debts to zero. This lets you fix situations where client bugs have resulted in distorted debt values. Set it, run the client, then clear it (otherwise you'll start out with zero debts every time you run the client). New in 6.6.11


Gruß,
Gundolf

Computer sind nicht alles im Leben. (Kleiner Scherz)

mikey
mikey
Joined: 22 Jan 05
Posts: 12706
Credit: 1839111349
RAC: 3589

RE: RE: einstein CPU

Quote:
Quote:
einstein CPU scheduling 1288
work fetch -37658

That's definitely a Long Term Debt issue.

If you don't agree with BOINC's assessment, you could zero that value for all projects by setting (temporarily) 1 in cc_config.xml.

Quote:
Set resource scheduling debts to zero. This lets you fix situations where client bugs have resulted in distorted debt values. Set it, run the client, then clear it (otherwise you'll start out with zero debts every time you run the client). New in 6.6.11

Gruß,
Gundolf

Or in the older versions of Boinc you could exit Boinc and manually edit the file and then restart Boinc and it will use the new settings. When I edit mine I just replace the numbers TO THE LEFT of the decimal point with a 1 and then leave the rest alone.

I do like the new settings though.

To explain a little about LTD, or Long Term Debt, it is used by Boinc the one on our pc's to keep track of your percentage settings for each project and try to keep them on track. So in your case you have Einstein at 80% and World Community Grid(WCG)at 10% and then Milky way at the other 10%. If Milky Way is using your gpu's only this does not count, they are done differently. Now to get back to the LTD....when Boinc contacts a project to get work and it doesn't have any it will go to one of your other projects to get it, it after is trying to maintain your cache of x amount of days of work on the pc. So if Einstein is out of the kind of work you want it will go to WCG and try to get some from them, while at the same time setting up some LTD, ie crunching time, for WCG so later on it will do more Einstein than WCG work. Right now your LTD for Einstein is HUGE, if you go into your cc_config.xml file and reset that to 1.000000 then it will ask for Einstein work again, because it is debt free. Otherwise over time Boinc WILL fix it on its own by your pc crunching more WCG units. The fix could take weeks or even months but Boinc WILL fix it if you just leave it alone! But most of us 'want it now' so we edit the file and poof it is fixed! IF you edit the file BE CAREFUL and only edit it in Notepad, NOT a word processing program!!!!! Word processing programs add junk to the file and it will NOT work right afterwards!! Be sure to exit Boinc before editing the file, save the file afterwards and then restart Boinc.

Ascholten
Ascholten
Joined: 20 Oct 10
Posts: 63
Credit: 2848108
RAC: 0

What if I just let Boinc on

What if I just let Boinc on this one system run out of all jobs, shut it down then restart with a fresh slate, will it 'fix itself' then possibly? It seems if there are no more jobs, then any 'backlog' should be null right?

Aaron

If god meant for us not to BOINC he'd have made our #$%^%^ shorter!!

Gundolf Jahn
Gundolf Jahn
Joined: 1 Mar 05
Posts: 1079
Credit: 341280
RAC: 0

RE: It seems if there are

Quote:
It seems if there are no more jobs, then any 'backlog' should be null right?


Nope, I think the LTD is carried forward even then.

mikey wrote:
if you go into your cc_config.xml file and reset that to 1.000000


That would be the client_state.xml file; very dangerous to edit!

Gruß,
Gundolf

Computer sind nicht alles im Leben. (Kleiner Scherz)

paul milton
paul milton
Joined: 16 Sep 05
Posts: 329
Credit: 35825044
RAC: 0

RE: What if I just let

Quote:

What if I just let Boinc on this one system run out of all jobs, shut it down then restart with a fresh slate, will it 'fix itself' then possibly? It seems if there are no more jobs, then any 'backlog' should be null right?

Aaron

not with out following the steps out lined above. boinc will still see one project has debt to another and only allow the one owed work.

example, 2 projects, A and B both at 50% but project A has been down for a week and couldnt get work, all the while project B has been crunchin along. after a week project A comes back up, boinc will now stop project B untill the 1 week debt has been paid back to project A.

the only way around it is to edit the debt values.

seeing without seeing is something the blind learn to do, and seeing beyond vision can be a gift.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.