Been crunching for 2 days and completed tasks showing credit claimed and credit granted as 0.00 - any suggestions?

Credit is granted based on the number of trickle_up files received.
It's possible that the trickle-to-credit script has failed to run, or that you need to wait a few more hours. (I think that it runs overnight, UK time.)

Suggestions? Yes! Keep your pants on! You completed a couple of short tasks mere hours ago. What do you want? Instantaneous credits posted? Not the way it's done here.

Relax. Work for the science. Ignore the 'glory.' Credits will be updated in due time -- whenever the script is run (perhaps in a day or three). It worked this way -- successfully -- for more than a decade.

Question: Are you running CPDN for the advancement of science? Or for credit?

[EDIT] Beat me to it again, Les!

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

Suggestions? Yes! Keep your pants on! You completed a couple of short tasks mere hours ago. What do you want? Instantaneous credits posted? Not the way it's done here.

Relax. Work for the science. Ignore the 'glory.' Credits will be updated in due time -- whenever the script is run (perhaps in a day or three). It worked this way -- successfully -- for more than a decade.

Question: Are you running CPDN for the advancement of science? Or for credit?

[EDIT] Beat me to it again, Les!

Wow, sorry to offend you, it was a simple question as i'm used to FAH. Promise I won't ask anymore questions here. The fact that the two tasks completed yesterday show 0.00 made me wonder if there is a problem with my system. What a great welcome to the forum reponse from a moderator !

Credits get asked about over and over, to the point where it starts getting tedious.
And this project is different to all others except QCN when it comes to credits.
Sometimes when there's been a problem in the past, it's taken days, and on occasions, weeks to get the credits working.

Also, each time the scripts run, they re-calculate EVERYTHING from scratch, including team credits. Which is why a team can lose a lot of credits if they lose a cruncher.

And, to add a bit more, pending credits here are an artefact of the BOINC system, and are never used.

Hi Currob,

Don't give up, it's worthwhile. Your efforts are really appreciated, especially with your impressive gear.
Even moderators have off days. Normally v helpful.

David

Been crunching for 2 days and completed tasks showing credit claimed and credit granted as 0.00 - any suggestions?

Yeah, sometimes it takes a while for credit to post here at CPDN.
Mostly cause the history is that, and even now, a WU on CPDN can be a week or a month or even more. They devised the "trickle" idea to get intermediate credit - but that doesn't always work.
So, unlike most other BOINC projects,
credit may take several days to show up. The whole project is on a longer timescale - WU-s run a few days to a few months. Nature of the beast.
That doesn't mean that useful uploads didn't happen, it's just that the credit system at CPDN is slow, and accurate.
I see you reported a few completed WU. Thanks. Keep crunching.
Credit will probably show up soon.
WU completed is the gold standard, I see you done that.

I've contributed a lot to CPDN - and I don't think efforts wasted, either credits or CPU time.

---

Thanks for the replies, I understand it better now, been folding for FAH for years and this is different, I reinstalled everything and leaving it running now, will leave this and one other pc running this project and others on the FAH project. I'm used to seeing credits (or points per day) change hour by hour on FAH and is great way to track my progress and also fine tune the system to get the most out of it. I gave some time to this project a year or two ago, wish I'd continued rather than put all my systems into one project.

This thread is interesting to me. About 4 days ago I set up a spare box with Ubuntu 15.04 (x86_64) and had a hard time getting the BOINC mgr to connect to the project. Then I had it offline while I was investigating potential problems with the computer's configuration. (BTW I had installed all the 32bit libs recommended, and more.)

Yesterday I put it back in service, and a miracle! It was running 4 CPDN tasks, around 5% complete! Excellent. However, it was troubling to later find no evidence of communication with the servers, no sign of trickle up has yet been seen.

This is in contrast to my Windows machine which very conscientiously uploads its data regularly. Believe me, I'm not worried about "credits", but Linux applications' patterns vs. Windows makes me a little nervous whether Linux apps are working correctly, that is, that the generated data is being sent to the project as it should be.

I'm guessing the Windows vs. Linux applications could be (or might very well have to be) programmed according to dissimilar OS requirements, hence wouldn't function exactly the same way. Nonetheless, reducing ambiguity re: what to expect when running tasks on Linux would be very helpful and appreciated.

Since it appears far fewer Linux cores are available to the project to run Linux-only models, it would be worthwhile to encourage volunteers to do so. That's a reason I decided to impress some surplus hardware into service--it looked like adding Linux clients would be useful. Possibly, I can help work out kinks, and if they are worked out, I have the capacity to put more Linux systems on line.

The frequency of trickles varies according to the model type. I was going to say that I hadn't noticed a problem with trickles recently but on this machine http://climateapps2.oerc.ox.ac.uk/cpdnboinc/result.php?resultid=18716754

It is sending trickles about every two days on the eu regional tasks. However after being away from the machine for two weeks, it lost connectivity with the router while I was away and on return average credit seems to have halved. This is probably just due to the credit script running once a day at night. I shall know the answer to that tomorrow morning.

By going to the page for a task you are running you can see the trickles that have been uploaded.

It is currently ploughing through the last 5 76MB zip files that were waiting to be uploaded.

There are 12 or 13 trickles for the MOSES II EU models. Thus, you will need to get to about 8.5% done to see the first trickle. If it goes much past that without a trickle, then there may be a problem.

Thanks for that info. The Linux machine running for a couple of days was showing tasks ~6.5% done this morning, and should reach the 8.5% point in a day or so. I'll keep a good thought it goes OK.

Again the difference with models running under Windows is notable. Likely enough, it's a source of confusion for people wanting to use OS alternatives.

It might be a good idea to "publish" such application specs in a particular place, that way users can know what to expect, saving everyone time and energy, and make the effort more enjoyable. An important bonus is that real problems would be more readily identified.

jrapdx

If you're talking about Computer 1373652, then it has way too little memory. What's there is barely enough to run the OS.
The climate models run here are the full blown supercomputer programs developed and run by the UK Met Office, and we recommend a minimum of 1 to 1.5 gigs of ram per core.

Your real problem may well be "disk thrashing", where the computer is spending most of it's time swapping data back and forth between the ram and the hard disk swap file. It could take months, maybe even a year to complete a model on it.

Les Bayliss

You are right about that computer--an old laptop obviously not up to the job. Wasn't such a good idea, I should give up on it. Hopefully stopping it isn't too big an issue since it was just starting the work anyway.

OTOH computer 1373243 has 16G ram and having no trouble with the 4 tasks it's running. The only question was communication with the server, but I think that's been addressed. Accordingly, since task progress is just over 7% now, I'm anticipating pretty soon some network activity will show up in the Event Log.

Just for the record, credit has now gone up again and average seems rather high now but I am sure it will settle down. As it took two days to go up rather than one after resetting router, I had wondered if there was a problem with the script batch file but if there was it is now sorted.

OTOH computer 1373243 has 16G ram and having no trouble with the 4 tasks it's running. The only question was communication with the server, but I think that's been addressed. Accordingly, since task progress is just over 7% now, I'm anticipating pretty soon some network activity will show up in the Event Log.

That PC has just uploaded trickles from 3 of the models. Not sure as to the timing of the credits script which runs once a day. Anyway, the credits will either show up today or tomorrow.

Thanks for the info. The credit showed up in the last couple of hours--I didn't see it earlier today. Good to know the connection to the servers is working, if subject to delays.

Seems the Linux-running models have a different trickle upload schedule vs. Windows. On Windows the uploads occur quite regularly and credits granted almost right away. Linux apps appear to accumulate bigger chunks, send less frequently and credits delayed for some longer interval.

I understand how there can be platform differences and that's OK. After more exposure to the "habits" of each system no doubt it will become much less a mystery.

I suspect that the real reason for the difference, is that the 2 OS now run different model types. These have different run times, due to their different modelling.

I think that for all models, trickles are created once a model month. So, as they take a different amount of time to complete a month, trickles will be created at different wall clock intervals for the 2 types.

I suspect that the real reason for the difference, is that the 2 OS now run different model types. These have different run times, due to their different modelling.

That's similar to my thoughts, the OS differences lead to variations in approach to programming models. Of course I can't say I know that for sure, just my hunch about it based on experience writing programs to run on Windows and unix, some things have to be done differently.

I think that for all models, trickles are created once a model month. So, as they take a different amount of time to complete a month, trickles will be created at different wall clock intervals for the 2 types.

Interestingly the once a month timing would seem to be reflected in the deadline for task completion. The longer ones are due up to ~one year after starting giving about 12 "monthly" trickles. That matches up pretty close to when the first trickle occurred at ~8.5% completion. The short tasks may get one or two trickles for their 2-3 month "lifetimes". Is this "correlation" simply coincidence?

The Windows tasks don't seem to follow the "monthly" rule, then again I do understand different models, different OS, etc., will just do things their own way.