Posts by Woodles

1) Message boards : Number crunching : To late to validate.. (Message 4485)
Posted 30 Jul 2018 by Woodles

There is enough time to finish a WU.

Which one? For long its 4 days.

I think four days is too short (or the WUs are too long)

I've had two WUs running on an i7 @ 3.6GHz (not a slow machine) that were rejected as taking too long. The host runs 24/7 and has all cores available to Boinc.

One task took 4.8 days and the other took 6.7 days.

http://srbase.my-firewall.org/sr5/results.php?hostid=3866&offset=0&show_names=0&state=5&appid=

I've also had eleven other tasks take over four days on the same host, luckily these got validated but I imagine some other cruncher got the resends and wasted their time trying to complete them in time.

I also won't be doing any more long tasks once the ones in progress are finished (if they do in time!)

2) Message boards : Number crunching : Not started by deadline - canceled (Message 2185)
Posted 15 Dec 2015 by Woodles

Yep,that's fixed it Reb, many thanks. I knew it would be something stupid that I'd done :)

3) Message boards : Number crunching : Not started by deadline - canceled (Message 2182)
Posted 14 Dec 2015 by Woodles

Hi Reb,

Well spotted, at this time, the host thinks the time is 20:48. But it also thinks it's Tuesday the 15th. I must have messed up when I set it originally.

Corrected now, I'll see how it goes.

4) Message boards : Number crunching : Not started by deadline - canceled (Message 2178)
Posted 14 Dec 2015 by Woodles

I've recently added a new host (3223) to the project and it now has 85,000+ tasks marked as "Not started by deadline - canceled"

For example: http://srbase.myfirewall.org/sr5/result.php?resultid=96766818

Sent: 14 Dec 2015, 14:13:45 UTC
Received 14 Dec 2015, 14:16:09 UTC
Report deadline 15 Dec 2015, 14:13:45 UTC
Returned 14 Dec 2015, 14:16:09 UTC - Not started by deadline - canceled

The host has successfully ran the tasks before (and still is now) but my other hosts don't have the problem so I assume it's some setting that I've got wrong.

The entry says "Aborted by user" but the host is remotely located and I haven't been near it for at least three days. Besides, the error code is "200 (0xc8) EXIT_UNSTARTED_LATE"

Settings for work fetch are currently at 0.1 days with 0.1 days extra and there's only 10 WUs in progress (it's an eight core machine) so I don't think the cache is too big for the WUs to get crunched before the deadline.

It's not a problem for me as the host quite happily crunches WU 24/7 but I don't know what effect it might be having on the project.

5) Message boards : Number crunching : SRBase not following sub project preferences (Message 2116)
Posted 27 Nov 2015 by Woodles

Hi Reb,

I've left it running for 24 hours and there's no incorrect WUs been sent. Looks like it's working as it should now.

I don't know what the problem was previously but it seems to have gone away. Thanks for your investigating and patience.

6) Message boards : Number crunching : SRBase not following sub project preferences (Message 2112)
Posted 26 Nov 2015 by Woodles

Hi Reb,

I have been crunching just one sub-project with the box unticked so I've changed it to multiple sub-projects (some of which don't have work) and ticked the box.

All working fine so far but I'll let you know if it goes wrong again.

Thanks for looking into this.

7) Message boards : Number crunching : SRBase not following sub project preferences (Message 2105)
Posted 25 Nov 2015 by Woodles

Hi Reb,

I have read all the posts again and try to figure out what is selected. So if you only select sub projects and "If no work for selected applications is available, accept work from other applications?" you will get work from non selected apps where work is available. If the last box is unselected and one sub project is enable but no work you will get no work at all, right? Or do you mean you will get also get work?

That's correct.

In my test I have enable "If no work for selected applications is available, accept work from other applications?" and one sub project where work is available and only getting work from this sub project. This is how it should be working. If you have the same then its ok.

That works for me too but only if the sub-project with work is the only one selected. If I also have another sub-project selected that doesn't have work then I get work from any sub-project with work.

To clarify (hopefully)

If I have just one sub-project selected which has work available then I get work for only that sub project whether I have the "accept work from other applications" option ticked or not - Correct operation.

If I have a selection of sub-projects selected that all have work then I get work for only those sub-projects whether I have the "accept work from other applications" option ticked or not - Correct operation.

If I have a selection of sub-projects selected, one or more of them don't have work available and I don't have the "accept work from other applications" option ticked then I only get work for the sub-projects that have work - Correct operation

If I have a selection of sub-projects selected, one or more of them doesn't have work available and I do have the "accept work from other applications" option ticked then I get work for any sub-projects that has work whether I've selected it or not - Incorrect operation.

It's almost as if the scheduler picks one of my selected sub-project then either sends work if there's some available for that sub-project or (if the option is selected) sends work from other sub-projects without checking to see if any other of my selected sub-project has work.

8) Message boards : Number crunching : SRBase not following sub project preferences (Message 2100)
Posted 24 Nov 2015 by Woodles

ok I see, I will enable debug logs later today to see where the problem comes from.

Thanks Reb, sorry to be a nuisance.

Should I re-enable other sub-projects for your testing?

9) Message boards : Number crunching : SRBase not following sub project preferences (Message 2098)
Posted 24 Nov 2015 by Woodles

Iam trying to keep constantly work for app 1,2,3,4,7,8 (one of them). You should enable all these apps + 1 average app. The runtimes for the short up to normal apps are 43s-12min on a 3.8Ghz AVX CPU.

But that's what I'm trying to say (badly!)

I had a selection of suitable WUs chosen, at least one of which always had work available but unless ALL of them had work, I got sent WUs from other sub-projects that I hadn't selected and couldn't work on.

For now, I'm only selecting one application at a time and not ticking the box. I'll stop crunching when the work for that sub-project runs out but other Boinc projects will pick up the slack until I notice. Not ideal but at least there won't be wasted work.

10) Message boards : Number crunching : SRBase not following sub project preferences (Message 2093)
Posted 23 Nov 2015 by Woodles

The selection box is labelled as "If no work for selected applications is available, accept work from other applications?"

ie. "If no work for selected applications is available ..."

I expected it to send work from other sub projects only if all the selected ones were without work.

So what you're saying is that if I select a few of the shorter sub projects (since the host goes off frequently) and any one of them doesn't have work then instead of switching to other WUs that I have selected to work on, I'll waste time crunching the longer sub projects that I have no chance of finishing?! Unless of course, I immediately change my preferences to exclude the sub project that's ran out of work! Or I could of course, only select one sub project at a time and not do any work at all when it runs out!

That might work if the sub projects had work for a decent length of time but with them changing nearly daily, it's going to mean almost hourly micro management.

Another solution would be to download a lot of work at a time and build up a stock of the work I want to do ... only SRBase sets a low limit on the amount of tasks you can stockpile.

I would like to add that I'm in no way criticising the project or the work that you've done, just trying to clarify how it works and find a way to get it to fit in with my requirements.

11) Message boards : Number crunching : SRBase not following sub project preferences (Message 2085)
Posted 20 Nov 2015 by Woodles

Hi Reb,

That's (almost) what I had. I had Reisel base short, Riesel base and Sierpinski / Riesel Base - average2 selected.

Reisel base short had work available, Riesel base and Sierpinski / Riesel Base - average2 didn't.

I had Sierpinski / Riesel Base - average3, Sierpinski / Riesel Base - average, Sierpinski / Riesel Base - long and Sierpinski Base - short all unselected.

I still got sent work for them.

Does the 'accept work from other projects' apply to each individual sub-project?

So if even one of the sub-projects I have selected has no work, I'll be sent random work from all the sub-projects regardless of whether any other of my selections have work?

If I have to uncheck a sub-project every time it runs out of work just to make sure that I get the work I selected then this project is going to take a lot of babysitting!

12) Message boards : Number crunching : SRBase not following sub project preferences (Message 2080)
Posted 19 Nov 2015 by Woodles

I have one of my hosts (#468) set to request only Riesel base, Riesel base short and Sierpinski / Riesel Base - average2 WUs. I also have the 'accept work from other projects if no work is available' box ticked.

Granted, two of these sub projects have no work at the moment but there are (currently) 304,270 Reisel base short WUs available.

My problem is that over the last couple of days, I have repeatedly been sent Sierpinski / Riesel Base - average3, Sierpinski / Riesel Base - average, Sierpinski / Riesel Base - long and Sierpinski Base - short WUs.

I can just about understand the Sierpinski / Riesel Base - averageN WUs being sent if the selection isn't quite specific enough but not the others.

Should I untick the 'accept work from other projects if no work is available' box or is there something else I should adjust? It's becoming a bit of a pain to continually check and abort WUs.