> I joined a few days ago. I've downloaded about 24 WU and uploaded about 16
> completed ones. What bothers me is that only one of those WUs have been sent
> to any other computers. Why is this?
>
> http://einsteinathome.org/account/tasks
Steve,
The locality scheduler, which decides what work to send to a given user, uses a random algorithm to assign data files (and hence the associated WU) to a given host. In your case, the file that you have has been assigned to at least one other user, who has picked up some of the WU. You'll find that within the next 2-3 days several other users will get the same data file and start doing the same WU as well.
Just be patient: other machines WILL get assigned the work, and you will get credit (lots of it!).
Ken;
Let your friend know that I just started a short time ago he is not the only one that has the same thing going on. I have 16 WUs that are pending and out of those 9 have not gone out to other computers and 1 only to one more. Let him know in time it will all catch up I'm sure. Here in mine if he just wants to look.
It takes time if you do 4 WUs a day for the program to catch up. Credits look nice because it shows to you that you are doing something and people know it, but not all computers are created the same. Things will catch up. Heck I have almost as much pending as I have been granted.
I'm the one with the pending credits Ken mentioned. What concerns me is that after a week the pending credits are still increasing faster than the granted credits. I have about 2100 pending and 433 granted.
> I'm the one with the pending credits Ken mentioned. What concerns me is that
> after a week the pending credits are still increasing faster than the granted
> credits. I have about 2100 pending and 433 granted.
If you "drill down" through your results listing you will see that many of your "pendings" were completed around 23,24,25,26, etc, Feb but were not issued to other crunchers until around 1,2 March. Also there are more recent "pendings" where yours is the only comp to have yet been given the work. Just the luck of the draw I guess. The money's safely in the Bank!! You'll get it all when the cheques clear!! :). I'd like to be in your position with all that pending loot to collect!!
I see Evan has posted about his situation. I'm very grateful to him for coming here. :)
First let me say, that Evan is certainly not out solely for credit: he has more than 1000 posts at the Bad Astronomy BB ("Bad" is used ironically there, as you know :).) It's just that he is determined to find out why his credit is so slow to come.
Last night after seeing his concern about the pending credit:
"I have 13 work units I have completed since and including Feb 26 which have not been sent to any other box,"
I thought, "that is a suspiciously long time," and I promised Evan that I would post the details for *all* his WU that had not been sent to another box to see what you would make of them.
I compiled this list before having read blerton's post, which I believe summarizes the situation the list portrays pretty well.
But a promise is a promise, and having done the list, I'm determined to post it!
And now the list of outstanding WUs that have not been sent to at least one other box, with "in the wild" meaning the approximate time since the WU was sent until 11h UTC 3/2 when I compiled the list:
*******
All later WUs are either in progress or completed on Evan's box, and have been sent to no other boxes. I hope I did the times right. :)
----------------------------
So if the window of "3-4" days is taken to mean that WUs must be farmed out within 4 days, then there is in fact only the 1 WU, 403939, in violation. If it is taken as 3 days, there are a total of 5 outside the range. As I said, late last night, it seemed there could be more; but OTOH, it is borderline for a few.
So anyway, I/we will keep and eye on these WUs and see when they clear.
I am very glad Evan is here to present his position, he does it very well. :) Also, I hope the E@H folks here understand that in presenting this list, I'm supporting my team member, and I'm not doing it to be critical of you.
> So if the window of "3-4" days is taken to mean that WUs must be farmed out
> within 4 days, then there is in fact only the 1 WU, 403939, in violation. If
> it is taken as 3 days, there are a total of 5 outside the range. As I said,
> late last night, it seemed there could be more; but OTOH, it is borderline for
> a few.
>
> So anyway, I/we will keep and eye on these WUs and see when they clear.
>
> I am very glad Evan is here to present his position, he does it very well. :)
> Also, I hope the E@H folks here understand that in presenting this list, I'm
> supporting my team member, and I'm not doing it to be critical of you.
>
> Thanks for putting up with me. :)
Ken,
I'm sure that nobody at EAH has even the slightest problem with any of the concerns you have raised on behalf of a team member. Prof Allen stated earlier in this thread in relation to a very similar situation being experienced by Steve Dyer:-
> The locality scheduler, which decides what work to send to a given user, uses
> a random algorithm to assign data files (and hence the associated WU) to a
> given host.
I took that to mean that because of the random behaviour, there would be cases where work was sent to one cruncher but not sent in a timely manner to other crunchers. Both Steve and Evan have obviously been victims of this. The problem then self corrects after a few days when the locality scheduler wakes up to itself and sends the work to other crunchers. Even if the expiry date is passed for the initial work before the subsequent work is processed, everything will still be fine in the end.
It's just an unfortunate side effect of how things work.
At least some of mine aren't getting credits because all the other machines that the wu has been sent to have timed out because the reporting limit is to short, but that's another thread - cough cough!
They will be sent out again one day, and eventually may get scrunched by the requisite number of clients, but I'm not holding my breath.
Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.
> At least some of mine aren't getting credits because all the other machines
> that the wu has been sent to have timed out because the reporting limit is to
> short, but that's another thread - cough cough!
Actually just 1 of yours is being delayed because all other crunchers have exceeded the time limit. If you investigate why, you find that (in one case anyway) the very first three units of one person all errored out the same way and then the comp started knocking over further units quite quickly. Sort of points towards some sort of configuration problem right in the beginning for that person. Whilst it's not nice to have your validation held up like this, it's not very fair to blame the 7 day limit. If the limit had been 14 days those other crunchers may have errored out at 14 days anyway and you'd be even worse off.
> They will be sent out again one day, and eventually may get scrunched by the
> requisite number of clients, but I'm not holding my breath.
Nor do you have to. The credit is in the Bank and eventually the funds will clear. Of your other 2 completed results, you already have credit for 1 and the other will validate very shortly if you look closely. You are actually in quite good shape compared to poor Steve and Evan.
blerton wrote:
>
> Ken,
>
> Prof Allen stated
> earlier in this thread in relation to a very similar situation being
> experienced by Steve Dyer:-
>
> > The locality scheduler, which decides what work to send to a given user,
> uses
> > a random algorithm to assign data files (and hence the associated WU) to
> a
> > given host.
>
> I took that to mean that because of the random behaviour, there would be cases
> where work was sent to one cruncher but not sent in a timely manner to other
> crunchers. Both Steve and Evan have obviously been victims of this. The
> problem then self corrects after a few days when the locality scheduler wakes
> up to itself and sends the work to other crunchers. Even if the expiry date
> is passed for the initial work before the subsequent work is processed,
> everything will still be fine in the end.
>
> It's just an unfortunate side effect of how things work.
>
Gary, thanks much!
I had indeed passed on to Evan Dr Allen's quote (and Dennis' answer in reply to Steve,) but it is understandably hard to get inside even wise words when you're watching your hard-crunched WUs seemingly march over the hill to 7th day doom. :)
Now, your last sentence quoted above and your reply to adrianxw:
>
>The credit is in the Bank and eventually the funds will clear.
>
implies to me that the 7 day deadline is really irrelevant: credit is given to all who have ever successfully returned a WU once some future gang of 3 confirms in its own 7 day window.
Do I have this right? If so, it's wonderful news for Evan's, Steve's and all fast boxes!
In fact, I just looked at my own list, and I remember having two red "No Reply" time outs; now there is only one.
If this is in the FAQ or earlier posts here, I apologize , but I'm glad to take my punishment, because it's such a great thing. :)
> I joined a few days ago.
)
> I joined a few days ago. I've downloaded about 24 WU and uploaded about 16
> completed ones. What bothers me is that only one of those WUs have been sent
> to any other computers. Why is this?
>
> http://einsteinathome.org/account/tasks
Steve,
The locality scheduler, which decides what work to send to a given user, uses a random algorithm to assign data files (and hence the associated WU) to a given host. In your case, the file that you have has been assigned to at least one other user, who has picked up some of the WU. You'll find that within the next 2-3 days several other users will get the same data file and start doing the same WU as well.
Just be patient: other machines WILL get assigned the work, and you will get credit (lots of it!).
Bruce
Director, Einstein@Home
Ken; Let your friend know
)
Ken;
Let your friend know that I just started a short time ago he is not the only one that has the same thing going on. I have 16 WUs that are pending and out of those 9 have not gone out to other computers and 1 only to one more. Let him know in time it will all catch up I'm sure. Here in mine if he just wants to look.
http://einstein.phys.uwm.edu/pending.php
It takes time if you do 4 WUs a day for the program to catch up. Credits look nice because it shows to you that you are doing something and people know it, but not all computers are created the same. Things will catch up. Heck I have almost as much pending as I have been granted.
John
When I first started running
)
When I first started running Einstein@Home my Pending Credits went up to 15,000 before it started to drop ... :)
I'm the one with the pending
)
I'm the one with the pending credits Ken mentioned. What concerns me is that after a week the pending credits are still increasing faster than the granted credits. I have about 2100 pending and 433 granted.
> I'm the one with the
)
> I'm the one with the pending credits Ken mentioned. What concerns me is that
> after a week the pending credits are still increasing faster than the granted
> credits. I have about 2100 pending and 433 granted.
If you "drill down" through your results listing you will see that many of your "pendings" were completed around 23,24,25,26, etc, Feb but were not issued to other crunchers until around 1,2 March. Also there are more recent "pendings" where yours is the only comp to have yet been given the work. Just the luck of the draw I guess. The money's safely in the Bank!! You'll get it all when the cheques clear!! :). I'd like to be in your position with all that pending loot to collect!!
Cheers,
Gary.
I see Evan has posted about
)
I see Evan has posted about his situation. I'm very grateful to him for coming here. :)
First let me say, that Evan is certainly not out solely for credit: he has more than 1000 posts at the Bad Astronomy BB ("Bad" is used ironically there, as you know :).) It's just that he is determined to find out why his credit is so slow to come.
Last night after seeing his concern about the pending credit:
"I have 13 work units I have completed since and including Feb 26 which have not been sent to any other box,"
I thought, "that is a suspiciously long time," and I promised Evan that I would post the details for *all* his WU that had not been sent to another box to see what you would make of them.
I compiled this list before having read blerton's post, which I believe summarizes the situation the list portrays pretty well.
But a promise is a promise, and having done the list, I'm determined to post it!
So first here's a link to Evan's Athlon:
http://einsteinathome.org/host/38890
And now the list of outstanding WUs that have not been sent to at least one other box, with "in the wild" meaning the approximate time since the WU was sent until 11h UTC 3/2 when I compiled the list:
(in order of oldest first)
http://einsteinathome.org/workunit/403854
Sent: 2/26; likely some kind of error since no box has finished it
http://einsteinathome.org/workunit/403939
sent 2/26; has been sent to no other box
in the wild 4d 0h approx
http://einsteinathome.org/workunit/405641
sent 2/26; has been sent to no other box
in the wild 3d 19h approx
http://einsteinathome.org/workunit/407037
sent: 2/26; likely some kind of error since no box has finished it
http://einsteinathome.org/workunit/407045
sent 2/26; has been sent to no other box
in the wild: 3d 17h approx
http://einsteinathome.org/workunit/409151
sent 2/27; has been sent to no other box
in the wild: 3d 6h approx
http://einsteinathome.org/workunit/410631
sent 2/27; has been sent to no other box
in the wild: 3d 0h approx
http://einsteinathome.org/workunit/412186
sent 2/27; has been sent to no other box
in the wild: 2d 18h approx
http://einsteinathome.org/workunit/413700
sent 2/27; has been sent to no other box
in the wild: 2d 13h approx
http://einsteinathome.org/workunit/417227
sent 2/28; has been sent to no other box
in the wild: 2d 0h approx
*******
All later WUs are either in progress or completed on Evan's box, and have been sent to no other boxes. I hope I did the times right. :)
----------------------------
So if the window of "3-4" days is taken to mean that WUs must be farmed out within 4 days, then there is in fact only the 1 WU, 403939, in violation. If it is taken as 3 days, there are a total of 5 outside the range. As I said, late last night, it seemed there could be more; but OTOH, it is borderline for a few.
So anyway, I/we will keep and eye on these WUs and see when they clear.
I am very glad Evan is here to present his position, he does it very well. :) Also, I hope the E@H folks here understand that in presenting this list, I'm supporting my team member, and I'm not doing it to be critical of you.
Thanks for putting up with me. :)
Ken
> So if the window of "3-4"
)
> So if the window of "3-4" days is taken to mean that WUs must be farmed out
> within 4 days, then there is in fact only the 1 WU, 403939, in violation. If
> it is taken as 3 days, there are a total of 5 outside the range. As I said,
> late last night, it seemed there could be more; but OTOH, it is borderline for
> a few.
>
> So anyway, I/we will keep and eye on these WUs and see when they clear.
>
> I am very glad Evan is here to present his position, he does it very well. :)
> Also, I hope the E@H folks here understand that in presenting this list, I'm
> supporting my team member, and I'm not doing it to be critical of you.
>
> Thanks for putting up with me. :)
Ken,
I'm sure that nobody at EAH has even the slightest problem with any of the concerns you have raised on behalf of a team member. Prof Allen stated earlier in this thread in relation to a very similar situation being experienced by Steve Dyer:-
> The locality scheduler, which decides what work to send to a given user, uses
> a random algorithm to assign data files (and hence the associated WU) to a
> given host.
I took that to mean that because of the random behaviour, there would be cases where work was sent to one cruncher but not sent in a timely manner to other crunchers. Both Steve and Evan have obviously been victims of this. The problem then self corrects after a few days when the locality scheduler wakes up to itself and sends the work to other crunchers. Even if the expiry date is passed for the initial work before the subsequent work is processed, everything will still be fine in the end.
It's just an unfortunate side effect of how things work.
Cheers,
Gary.
At least some of mine aren't
)
At least some of mine aren't getting credits because all the other machines that the wu has been sent to have timed out because the reporting limit is to short, but that's another thread - cough cough!
They will be sent out again one day, and eventually may get scrunched by the requisite number of clients, but I'm not holding my breath.
Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.
> At least some of mine
)
> At least some of mine aren't getting credits because all the other machines
> that the wu has been sent to have timed out because the reporting limit is to
> short, but that's another thread - cough cough!
Actually just 1 of yours is being delayed because all other crunchers have exceeded the time limit. If you investigate why, you find that (in one case anyway) the very first three units of one person all errored out the same way and then the comp started knocking over further units quite quickly. Sort of points towards some sort of configuration problem right in the beginning for that person. Whilst it's not nice to have your validation held up like this, it's not very fair to blame the 7 day limit. If the limit had been 14 days those other crunchers may have errored out at 14 days anyway and you'd be even worse off.
> They will be sent out again one day, and eventually may get scrunched by the
> requisite number of clients, but I'm not holding my breath.
Nor do you have to. The credit is in the Bank and eventually the funds will clear. Of your other 2 completed results, you already have credit for 1 and the other will validate very shortly if you look closely. You are actually in quite good shape compared to poor Steve and Evan.
Cheers,
Gary.
blerton wrote: > > Ken, > >
)
blerton wrote:
>
> Ken,
>
> Prof Allen stated
> earlier in this thread in relation to a very similar situation being
> experienced by Steve Dyer:-
>
> > The locality scheduler, which decides what work to send to a given user,
> uses
> > a random algorithm to assign data files (and hence the associated WU) to
> a
> > given host.
>
> I took that to mean that because of the random behaviour, there would be cases
> where work was sent to one cruncher but not sent in a timely manner to other
> crunchers. Both Steve and Evan have obviously been victims of this. The
> problem then self corrects after a few days when the locality scheduler wakes
> up to itself and sends the work to other crunchers. Even if the expiry date
> is passed for the initial work before the subsequent work is processed,
> everything will still be fine in the end.
>
> It's just an unfortunate side effect of how things work.
>
Gary, thanks much!
I had indeed passed on to Evan Dr Allen's quote (and Dennis' answer in reply to Steve,) but it is understandably hard to get inside even wise words when you're watching your hard-crunched WUs seemingly march over the hill to 7th day doom. :)
Now, your last sentence quoted above and your reply to adrianxw:
>
>The credit is in the Bank and eventually the funds will clear.
>
implies to me that the 7 day deadline is really irrelevant: credit is given to all who have ever successfully returned a WU once some future gang of 3 confirms in its own 7 day window.
Do I have this right? If so, it's wonderful news for Evan's, Steve's and all fast boxes!
In fact, I just looked at my own list, and I remember having two red "No Reply" time outs; now there is only one.
If this is in the FAQ or earlier posts here, I apologize , but I'm glad to take my punishment, because it's such a great thing. :)
Ken