Deferring communication with project?

rygraham
rygraham
Joined: 26 Feb 05
Posts: 4
Credit: 126205
RAC: 0
Topic 188099

I completed a WU and it uploaded, but BOINC is still deferring communication with project (on a 24 hour schedule). Shouldn't a new work unit download upon completion? BOINC is idle right now.

mlcudd
mlcudd
Joined: 22 Jan 05
Posts: 26
Credit: 51808
RAC: 0

Deferring communication with project?

What do you have set in your preferences for time span to connect to server? If this is 0 you will not get new workunits...just a thought. I have mine set for 3 days. The server page shows everything up, and WU's to send.

Regards,

Rocky


WinXP SP2 Boinc - 4.45
Have A Great Day!

rygraham
rygraham
Joined: 26 Feb 05
Posts: 4
Credit: 126205
RAC: 0

That was it -- it was set for

That was it -- it was set for network communication every 0.1 days. I set it to every 3 days, and it downloaded enough work units to fill the time.

Thanks!

marvinvwinkle
marvinvwinkle
Joined: 24 Feb 05
Posts: 1444
Credit: 85585
RAC: 0

I have the same problem and

I have the same problem and not being very smart mine is set for 0.1 and is deferring for 24 hours. Do you set it for 0.3 or 3.0 or 3? Help! Thanks

TFFE

Jack Brady
Jack Brady
Joined: 20 Feb 05
Posts: 4
Credit: 597
RAC: 0

> I completed a WU and it

> I completed a WU and it uploaded, but BOINC is still deferring communication
> with project (on a 24 hour schedule). Shouldn't a new work unit download upon
> completion? BOINC is idle right now.
> I got the same problem. I hope we get an answer. If not I will delete Boinc program. I would think you would be able to force a new work-unit load.

Darren
Darren
Joined: 18 Jan 05
Posts: 94
Credit: 69632
RAC: 0

What message are you getting

What message are you getting when it first starts the 24-hour deferral?

Depending on which OS you're using, there are ways to get past the deferral, but you need to post the part of the log showing why it made the deferral in the first place so someone can help you get that problem corrected first - otherwise it will just keep on deferring.

rygraham
rygraham
Joined: 26 Feb 05
Posts: 4
Credit: 126205
RAC: 0

> I have the same problem and

Message 6025 in response to message 6022

> I have the same problem and not being very smart mine is set for 0.1 and is
> deferring for 24 hours. Do you set it for 0.3 or 3.0 or 3? Help! Thanks
>

In the general preferences I set the "Connect to network about every" value to 3 (days). It then downloaded 11 work units to my processing queue. I get about 2 WU's done per day, so this should last at least 5 days.

HTH,
RG

Gary Roberts
Gary Roberts
Moderator
Joined: 9 Feb 05
Posts: 5887
Credit: 119308876325
RAC: 25562408

> In the general preferences

Message 6026 in response to message 6025

> In the general preferences I set the "Connect to network about every" value to
> 3 (days). It then downloaded 11 work units to my processing queue. I get
> about 2 WU's done per day, so this should last at least 5 days.
>
> HTH,
> RG

If you are doing 2 per day, by the time you get to #11 it will already be 5.5 days stale and although it will finish and be perfectly acceptable at 6 days, there will have been coworkers wondering why they have results sitting for many days in a pending queue waiting for your result. Also you face the risk that if your computer gets delayed for a day or two for some reason, that WU will expire before you get a chance to complete it. Also once you start to maintain a 5+ day cache, every new WU you add will be quite stale by the time you actually get to start work on it.

If you extrapolate this and large numbers of workers decide to keep large caches, the pending results queues could get quite large. Also the number of expiring WUs would increase dramatically due to relatively minor glitches tipping them over the 7 day limit. Sounds like a recipe for resource wastage if this went to the extreme.

Please don't take any of this as personal criticism. It's just a plea for a more moderate cache limit. Sure, servers go down and problems lasting many hours do occur so it is wise to have some cached work. In your case, wouldn't 4 or 5 or 6 units be sufficient giving you 2+ days of work. Hopefully it would be quite rare for that length of outage to occur. In any case you could always temporarily adopt another Boinc project if the worst came to the worst :).

Personally, I've adopted a policy of keeping about 12-24 hours of work on hand. I'm getting very fresh work, fresh timely results and a growing pending queue waiting for others to complete. I don't mind that at all but I'd probably start to get a little restive if every single pending result had to wait at least 5 days or more for validation. On second thoughts not even that would bother me but rather the thought of increasing wastage of precious resources does.

Cheers,
Gary.

Cheers,
Gary.

rygraham
rygraham
Joined: 26 Feb 05
Posts: 4
Credit: 126205
RAC: 0

> > In the general

Message 6027 in response to message 6026

> > In the general preferences I set the "Connect to network about every"
> value to
> > 3 (days). It then downloaded 11 work units to my processing queue. I
> get
> > about 2 WU's done per day, so this should last at least 5 days.
> >
> > HTH,
> > RG
>
> If you are doing 2 per day, by the time you get to #11 it will already be 5.5
> days stale and although it will finish and be perfectly acceptable at 6 days,
> there will have been coworkers wondering why they have results sitting for
> many days in a pending queue waiting for your result. Also you face the risk
> that if your computer gets delayed for a day or two for some reason, that WU
> will expire before you get a chance to complete it. Also once you start to
> maintain a 5+ day cache, every new WU you add will be quite stale by the time
> you actually get to start work on it.
>
> If you extrapolate this and large numbers of workers decide to keep large
> caches, the pending results queues could get quite large. Also the number of
> expiring WUs would increase dramatically due to relatively minor glitches
> tipping them over the 7 day limit. Sounds like a recipe for resource wastage
> if this went to the extreme.
>
> Please don't take any of this as personal criticism. It's just a plea for a
> more moderate cache limit. Sure, servers go down and problems lasting many
> hours do occur so it is wise to have some cached work. In your case, wouldn't
> 4 or 5 or 6 units be sufficient giving you 2+ days of work. Hopefully it
> would be quite rare for that length of outage to occur. In any case you could
> always temporarily adopt another Boinc project if the worst came to the worst
> :).
>
> Personally, I've adopted a policy of keeping about 12-24 hours of work on
> hand. I'm getting very fresh work, fresh timely results and a growing pending
> queue waiting for others to complete. I don't mind that at all but I'd
> probably start to get a little restive if every single pending result had to
> wait at least 5 days or more for validation. On second thoughts not even that
> would bother me but rather the thought of increasing wastage of precious
> resources does.
>
> Cheers,
> Gary.
>

This makes sense to me. I'll try dropping the setting from 3 days to 1 or 1.5 days for communication and see how that affects the number of WUs downloaded.

My machine is a dual processor, so I think BOINC downloaded enough WUs for both procs, but one is constantly cranking on climateprediciton.net. I wonder if I can alter this behavior by changing the resource allocation settings?

Thanks for the tip,
-Ryan

Comment viewing options

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