Why pending credit takes so long?

Message boards : Number crunching : Why pending credit takes so long?
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · Next

AuthorMessage
LGM

Send message
Joined: 10 Oct 07
Posts: 5
Credit: 6,640
RAC: 0
United States
Message 664382 - Posted: 22 Oct 2007, 13:52:26 UTC

I'm sure this has been covered in the past, so I
apologize for any redundancy. Most of my pending
credits go through quickly but I have this one task
that has been pending for days now. I was just
wondering the usual reasons for when this happens.

Thanks in advance,

LGM
ID: 664382 · Report as offensive
Profile purplemkayel
Avatar

Send message
Joined: 23 Jul 02
Posts: 1904
Credit: 55,594
RAC: 0
Australia
Message 664387 - Posted: 22 Oct 2007, 14:02:07 UTC

Welcome to the SETI forums,
Depending on your wingman, credit can take more than a month to arrive. For example, I'm waiting on credit from September... It'll arrive eventually :)
Happy birthday Calm Chaos!!! Terrible twos?


Calm Chaos... are you feeling it yet?
ID: 664387 · Report as offensive
Profile John Clark
Volunteer tester
Avatar

Send message
Joined: 29 Sep 99
Posts: 16515
Credit: 4,418,829
RAC: 0
United Kingdom
Message 664394 - Posted: 22 Oct 2007, 14:24:00 UTC

My wingman on this WU is so far behind me that I am beginning to think they have left S@H but not told the servers.

I completed the WU on August 27th, and still wait my wingman to report!

My claimed credit is too high by, I suspect, about 20 cobblestones.
It's good to be back amongst friends and colleagues



ID: 664394 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14649
Credit: 200,643,578
RAC: 874
United Kingdom
Message 664407 - Posted: 22 Oct 2007, 14:44:10 UTC

I'd been thinking of writing up a reply along these lines for a few days, and your question seems as good an excuse for posting it as any, so here goes.

At SETI, you can't earn credit by yourself. You have to have a second user, a 'wingman', who crunches the same result and gets near enough the same answer: then you validate each other's work and both get credit.

While only one result has been returned, the credit you have claimed shows as 'pending'. Sometimes it seems to stay that way for ages. Why? I can think of four different reasons.

1) Your wingman is slower than you are
In the sense that they take longer to return a result that's been issued to them than you do. They may indeed have a much slower computer: they may take part in a much larger number of BOINC projects than you do, and have to spread their computer power more thinly: they may only have their computer turned on for a small proportion of each day: or they may have opted to 'cache' a larger number of results than you do, so you have to wait while the other work is finished before they get to the result you're interested in. All perfectly reasonable choices, and just some of the different ways that BOINC is designed to be used. You just have to wait until their result comes in: it shouldn't be more than a few days.

2) Your wingman has gone missing in action
Some people sign up to BOINC and/or SETI, download one or two workunits, and then decide the project doesn't suit them and disappear without a trace. Others do a substantial amount of work, and download hundreds or even thousands of WUs, and then disappear equally suddenly. And it's always possible to suffer a computer breakdown, and not be able to return the work you really intended to finish. When this happens, the BOINC servers wait until the 'deadline' set when the task was issued, and then send another copy to someone else so that the comparison result can be finished off. At the moment, the SETI deadlines are quite long (up to three months), so you can be waiting for this long - plus whatever time your new wingman takes - before your pending credit is granted, but you'll get it in the end.

3) You and your wingman disagree about the result
If the results for the two tasks that SETI sends out are different, the servers send out a third copy. When this result comes back in (subject to any of the delays I've described already), the results are checked again. Sometimes all three are 'close enough', and all are awarded credit: sometimes two of them agree (and are awarded credit), but the third is adjudged faulty and gets nothing. If you get the credit, fine: but if yours is the one that gets thrown out, you should investigate further and take remedial action. Is your computer running well (and not overheating)? Are you running the right software? Could you have any faulty (or marginal) hardware components? Are all your software drivers (especially graphics card drivers) up to date?

4) You and you wingman are both fine, but SETI has forgotten to check in your result
Not common, but it does happen - I've just posted about it in another thread. All you can do is to alert people to the problem by posting in this message board, and hope that someone with the power to fix it reads your plea.
ID: 664407 · Report as offensive
MarkTW

Send message
Joined: 12 Aug 02
Posts: 1
Credit: 10,366,200
RAC: 17
United States
Message 664425 - Posted: 22 Oct 2007, 15:09:35 UTC

I have over 1,000 "credits" pending, which represents nearly a week of production for me. This has been building up for a couple of weeks, I think. Before this, my Pending level has rarely been highr than 200 or 300. Checking the details, I notice that the deadlines for many of the "wingmen" are well into December. Though many pending credits are resolved within a few days, quite a few take the full alotted time, expire, and are then passed on to a new "wingman." This can, and I think does, stretch out some results for a very long time. This has to add to the overhead of the system, maintaining records for extended periods, right? I'm sure it has beeen asked before, but I'm going to ask again, why doesn't the system allocate work units taking the turnaround performance of the computers into consideration? My systems generally turn units around within a couple of days. But I find I'm paired with computers that show turnaround times of more than a week, even as much as 3 weeks. It's irritating even though I know eventually everything will be cleared.
ID: 664425 · Report as offensive
Profile jedimstr
Avatar

Send message
Joined: 23 Oct 00
Posts: 33
Credit: 16,828,887
RAC: 0
United States
Message 664434 - Posted: 22 Oct 2007, 15:27:21 UTC - in response to Message 664425.  

I have over 1,000 "credits" pending, which represents nearly a week of production for me... snip...


Man, I'd be happy with 1,000 credits pending. Here's mine:
Pending credit: 30,866.48


ID: 664434 · Report as offensive
Profile perryjay
Volunteer tester
Avatar

Send message
Joined: 20 Aug 02
Posts: 3377
Credit: 20,676,751
RAC: 0
United States
Message 664438 - Posted: 22 Oct 2007, 15:35:05 UTC - in response to Message 664407.  

I'd been thinking of writing up a reply along these lines for a few days, and your question seems as good an excuse for posting it as any, so here goes.

At SETI, you can't earn credit by yourself. You have to have a second user, a 'wingman', who crunches the same result and gets near enough the same answer: then you validate each other's work and both get credit.

While only one result has been returned, the credit you have claimed shows as 'pending'. Sometimes it seems to stay that way for ages. Why? I can think of four different reasons.

1) Your wingman is slower than you are
In the sense that they take longer to return a result that's been issued to them than you do. They may indeed have a much slower computer: they may take part in a much larger number of BOINC projects than you do, and have to spread their computer power more thinly: they may only have their computer turned on for a small proportion of each day: or they may have opted to 'cache' a larger number of results than you do, so you have to wait while the other work is finished before they get to the result you're interested in. All perfectly reasonable choices, and just some of the different ways that BOINC is designed to be used. You just have to wait until their result comes in: it shouldn't be more than a few days.

2) Your wingman has gone missing in action
Some people sign up to BOINC and/or SETI, download one or two workunits, and then decide the project doesn't suit them and disappear without a trace. Others do a substantial amount of work, and download hundreds or even thousands of WUs, and then disappear equally suddenly. And it's always possible to suffer a computer breakdown, and not be able to return the work you really intended to finish. When this happens, the BOINC servers wait until the 'deadline' set when the task was issued, and then send another copy to someone else so that the comparison result can be finished off. At the moment, the SETI deadlines are quite long (up to three months), so you can be waiting for this long - plus whatever time your new wingman takes - before your pending credit is granted, but you'll get it in the end.

3) You and your wingman disagree about the result
If the results for the two tasks that SETI sends out are different, the servers send out a third copy. When this result comes back in (subject to any of the delays I've described already), the results are checked again. Sometimes all three are 'close enough', and all are awarded credit: sometimes two of them agree (and are awarded credit), but the third is adjudged faulty and gets nothing. If you get the credit, fine: but if yours is the one that gets thrown out, you should investigate further and take remedial action. Is your computer running well (and not overheating)? Are you running the right software? Could you have any faulty (or marginal) hardware components? Are all your software drivers (especially graphics card drivers) up to date?

4) You and you wingman are both fine, but SETI has forgotten to check in your result
Not common, but it does happen - I've just posted about it in another thread. All you can do is to alert people to the problem by posting in this message board, and hope that someone with the power to fix it reads your plea.



Thanks Richard, I hope everyone sees this post. It really explains all the details. I think I've had all of those except the last one. :)


PROUD MEMBER OF Team Starfire World BOINC
ID: 664438 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 19012
Credit: 40,757,560
RAC: 67
United Kingdom
Message 664450 - Posted: 22 Oct 2007, 15:57:04 UTC
Last modified: 22 Oct 2007, 15:59:28 UTC

For a short period before Multi-Beam was released the initial replication was three. This meant you had two wingmmen for each unit, but the min quorum was two, so if 50% of these wingmen reported quickly you got about 90% of credits granted quickly, 2 to 3 days.
You were happy because less than 10% of units went into the pending tray, probably equaling about half of your RAC score. And you only paid scant attention to the other 50% of your wingmen.

With MB the replication decreased to two. So now you only have 1 wingman, but these, on average, are of similar performance as before, some fast reports some slow. This still means only about 50% of wingmen will report within 3 days. Therefore only 50% of units, not 90%, will be granted credit quickly. The result lots more pending, for longer periods. Probably equal to 3 or even 4 * your RAC. Now all wingmen are important to you.
ID: 664450 · Report as offensive
1mp0£173
Volunteer tester

Send message
Joined: 3 Apr 99
Posts: 8423
Credit: 356,897
RAC: 0
United States
Message 664474 - Posted: 22 Oct 2007, 16:53:38 UTC - in response to Message 664425.  

I have over 1,000 "credits" pending, which represents nearly a week of production for me. This has been building up for a couple of weeks, I think. Before this, my Pending level has rarely been highr than 200 or 300. Checking the details, I notice that the deadlines for many of the "wingmen" are well into December. Though many pending credits are resolved within a few days, quite a few take the full alotted time, expire, and are then passed on to a new "wingman." This can, and I think does, stretch out some results for a very long time. This has to add to the overhead of the system, maintaining records for extended periods, right? I'm sure it has beeen asked before, but I'm going to ask again, why doesn't the system allocate work units taking the turnaround performance of the computers into consideration? My systems generally turn units around within a couple of days. But I find I'm paired with computers that show turnaround times of more than a week, even as much as 3 weeks. It's irritating even though I know eventually everything will be cleared.

Remember that BOINC is trying to predict the future.

You have a turn-around time of a couple of days. Great.

So, (hypothetically) you've got a full cache, deadlines comfortably in the future, but you're going on vacation, and for safety you shut down for a week.

... or the machine dies, and work goes with it.

... or a glitch resets the DCF to something low and the machine downloads TONS of work.

This is volunteer computing, and it is very difficult to predict what the volunteers are going to do.
ID: 664474 · Report as offensive
Profile perryjay
Volunteer tester
Avatar

Send message
Joined: 20 Aug 02
Posts: 3377
Credit: 20,676,751
RAC: 0
United States
Message 664476 - Posted: 22 Oct 2007, 16:57:09 UTC - in response to Message 664450.  

For a short period before Multi-Beam was released the initial replication was three. This meant you had two wingmmen for each unit, but the min quorum was two, so if 50% of these wingmen reported quickly you got about 90% of credits granted quickly, 2 to 3 days.
You were happy because less than 10% of units went into the pending tray, probably equaling about half of your RAC score. And you only paid scant attention to the other 50% of your wingmen.

With MB the replication decreased to two. So now you only have 1 wingman, but these, on average, are of similar performance as before, some fast reports some slow. This still means only about 50% of wingmen will report within 3 days. Therefore only 50% of units, not 90%, will be granted credit quickly. The result lots more pending, for longer periods. Probably equal to 3 or even 4 * your RAC. Now all wingmen are important to you.



I liked to watch and /or run in the race for second place. It was fun watching to see who got bumped out of the credits. :)



PROUD MEMBER OF Team Starfire World BOINC
ID: 664476 · Report as offensive
dblEagle
Avatar

Send message
Joined: 3 Apr 99
Posts: 136
Credit: 45,641
RAC: 0
United States
Message 664479 - Posted: 22 Oct 2007, 17:06:39 UTC

I think the main reason for the pending credits is because a lot of people haven't switched over to the new Core2 DUO's yet, including me. The difference in computing power is widening. The new Core2 DUO's are about twice as fast and are doing about twice the number of WU's. It takes a slower computer a little while to catch up.
ID: 664479 · Report as offensive
Profile Dr. C.E.T.I.
Avatar

Send message
Joined: 29 Feb 00
Posts: 16019
Credit: 794,685
RAC: 0
United States
Message 664499 - Posted: 22 Oct 2007, 17:49:37 UTC - in response to Message 664479.  

I think the main reason for the pending credits is because a lot of people haven't switched over to the new Core2 DUO's yet, including me. The difference in computing power is widening. The new Core2 DUO's are about twice as fast and are doing about twice the number of WU's. It takes a slower computer a little while to catch up.


and right ya be too . . . most people have to simply just wait for the return - NO biggie - this isn't a race . . . ;)))

runs for cover ;)


BOINC Wiki . . .

Science Status Page . . .
ID: 664499 · Report as offensive
Profile perryjay
Volunteer tester
Avatar

Send message
Joined: 20 Aug 02
Posts: 3377
Credit: 20,676,751
RAC: 0
United States
Message 664501 - Posted: 22 Oct 2007, 17:51:11 UTC - in response to Message 664479.  

I think the main reason for the pending credits is because a lot of people haven't switched over to the new Core2 DUO's yet, including me. The difference in computing power is widening. The new Core2 DUO's are about twice as fast and are doing about twice the number of WU's. It takes a slower computer a little while to catch up.


Nice thought dbleagle but all 4 of my machines are old and slow but I find myself waiting for core 2 and quads a lot. It seems the faster the machine they have they think they need a larger cache to keep it fed. I run through WUs long before they get to them Through their long cache.




PROUD MEMBER OF Team Starfire World BOINC
ID: 664501 · Report as offensive
Profile Clyde C. Phillips, III

Send message
Joined: 2 Aug 00
Posts: 1851
Credit: 5,955,047
RAC: 0
United States
Message 664511 - Posted: 22 Oct 2007, 18:24:53 UTC - in response to Message 664434.  

I have over 1,000 "credits" pending, which represents nearly a week of production for me... snip...


Man, I'd be happy with 1,000 credits pending. Here's mine:
Pending credit: 30,866.48



Jedi, I have about 1,650 RAC and about 4,000 pendings. That's a little over two daysworth of pendings. You have a similar proportion. I've had around 4,000 pendings for some time. So in your case, it looks like everything is AOK. Hell will freeze over before your pendings will drop to 1,000 (unless the system is changed). For that, mine would have to drop to about 135.

ID: 664511 · Report as offensive
Profile Andy Lee Robinson
Avatar

Send message
Joined: 8 Dec 05
Posts: 630
Credit: 59,973,836
RAC: 0
Hungary
Message 664552 - Posted: 22 Oct 2007, 19:59:46 UTC - in response to Message 664479.  
Last modified: 22 Oct 2007, 20:00:25 UTC

I think the main reason for the pending credits is because a lot of people haven't switched over to the new Core2 DUO's yet, including me. The difference in computing power is widening. The new Core2 DUO's are about twice as fast and are doing about twice the number of WU's. It takes a slower computer a little while to catch up.


The huge increase in pendings has absolutely NOTHING to do with the speed of a computer.

It has everything to do with the change to 2 instead of 3 to reach quorum, large caches and machines going AWOL for various reasons without any proactive method of reallocating their trapped workunits.

A slow computer with a 1 day cache can receive, process *and* return a few results a day.
The fastest computer with a 10 day cache will take 10 days to turn around the same workunit if sent at the same time.

SETI@home system can't process the workunit further until both crunchers have returned a successful result, so the database fills up with pending workunits.
One that is done, then the result can be saved, credit granted and the workunit deleted to make space - workunits are a third of a megabyte each so this is a subtantial amount of data to have lying around unnecessarily.

The most effective cache is the *smallest* necessary to ride outages. Big caches are for those few that really do have intermittent access.

SETI@Home will run much nicer and smoother if work is turned around quickly with the least amount of work remaining in circulation, and everyone will get credits more quickly.

Is that clear now?
ID: 664552 · Report as offensive
Osiris30

Send message
Joined: 19 Aug 07
Posts: 264
Credit: 41,917,631
RAC: 0
Barbados
Message 664564 - Posted: 22 Oct 2007, 20:16:13 UTC - in response to Message 664434.  

I have over 1,000 "credits" pending, which represents nearly a week of production for me... snip...


Man, I'd be happy with 1,000 credits pending. Here's mine:
Pending credit: 30,866.48



Tradesies:
Pending credit: 140,738.36
ID: 664564 · Report as offensive
Profile Philadelphia
Volunteer tester
Avatar

Send message
Joined: 12 Feb 07
Posts: 1590
Credit: 399,688
RAC: 0
United States
Message 664576 - Posted: 22 Oct 2007, 20:32:07 UTC - in response to Message 664564.  

I have over 1,000 "credits" pending, which represents nearly a week of production for me... snip...


Man, I'd be happy with 1,000 credits pending. Here's mine:
Pending credit: 30,866.48



Tradesies:
Pending credit: 140,738.36


Hmmm, three of you mentioned your pending number and I did a rough calculation to come up with the following:

Jedi Mstr - Pending credit: 30,866.48 - RAC: 13,834 - RAC % of PC: 44.8
Clyde C Phillips - PC: ~4,000 - RAC: 1,650 - RAC % of PC: 41.3
Osiris30 - PC: 140,738.36 - RAC: 60,355 - RAC % of PC: 42.9

I wonder if that trend of low 40% is the same for most others too? Probably not but I thought it was interesting that it was true for those three.



ID: 664576 · Report as offensive
Profile Andy Lee Robinson
Avatar

Send message
Joined: 8 Dec 05
Posts: 630
Credit: 59,973,836
RAC: 0
Hungary
Message 664581 - Posted: 22 Oct 2007, 20:40:06 UTC - in response to Message 664576.  
Last modified: 22 Oct 2007, 20:44:27 UTC

I wonder if that trend of low 40% is the same for most others too? Probably not but I thought it was interesting that it was true for those three.


26,649 here, 32.5% (cache 1+1 day).
ID: 664581 · Report as offensive
Profile dnolan
Avatar

Send message
Joined: 30 Aug 01
Posts: 1228
Credit: 47,779,411
RAC: 32
United States
Message 664583 - Posted: 22 Oct 2007, 20:42:59 UTC - in response to Message 664576.  
Last modified: 22 Oct 2007, 20:43:15 UTC



Hmmm, three of you mentioned your pending number and I did a rough calculation to come up with the following:

Jedi Mstr - Pending credit: 30,866.48 - RAC: 13,834 - RAC % of PC: 44.8
Clyde C Phillips - PC: ~4,000 - RAC: 1,650 - RAC % of PC: 41.3
Osiris30 - PC: 140,738.36 - RAC: 60,355 - RAC % of PC: 42.9

I wonder if that trend of low 40% is the same for most others too? Probably not but I thought it was interesting that it was true for those three.




Doesn't look like it for me, pending is 25491 as of right now, RAC 13395, so that would be something like 52 - 53%.

-Dave


ID: 664583 · Report as offensive
Profile [KWSN]John Galt 007
Volunteer tester
Avatar

Send message
Joined: 9 Nov 99
Posts: 2444
Credit: 25,086,197
RAC: 0
United States
Message 664590 - Posted: 22 Oct 2007, 20:57:00 UTC

Pending:20429
RAC:9214

About 45% for me.
Clk2HlpSetiCty:::PayIt4ward

ID: 664590 · Report as offensive
1 · 2 · 3 · Next

Message boards : Number crunching : Why pending credit takes so long?


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.