> I missed it again.
> My result was uploaded toonight within the deadline but when i wake up in the
> morning reported too late only a few minutes.
>
I have a question along these lines. I understand that uploading and reporting is a two-step process... but what i don't understand is why it is a two-step process. Is there no way that results could be reported at the same time they are uploaded?
It seems quite a shame that Mike completed the workunit on time, but because the reporting process is delayed, he lost his credit.
> I have a question along these lines. I understand that uploading and
> reporting is a two-step process... but what i don't understand is why
> it is a two-step process. Is there no way that results could be reported at
> the same time they are uploaded?
>
> It seems quite a shame that Mike completed the workunit on time, but because
> the reporting process is delayed, he lost his credit.
>
> Just a question, not a complaint. :)
I thought I had explained that ... but could only find mention that it is a two step process.
> I have a question along these lines. I understand that uploading and
> reporting is a two-step process... but what i don't understand is why
> it is a two-step process. Is there no way that results could be reported at
> the same time they are uploaded?
Why? - Because it saves the overhead of server contact. For immediate reporting:-
1. Upload result to data server
2. Report claim for credit to scheduling server
and sometime later...
3. Request more work from scheduling server.
For delayed reporting we only need:-
1. Upload result to data server
and sometime later...
2. Report claim for credit and request more work from scheduler.
> It seems quite a shame that Mike completed the workunit on time, but because
> the reporting process is delayed, he lost his credit.
Yes, it is a great shame. When I first started with one machine, I thought it would be a good idea (Seti classic mentality) to keep a goodly cache of work on hand, just in case.... I was about to go away on a one week business trip so I thought I'd "stock up". I had about 10 units on hand. Just before heading off to catch the plane, I changed my mind when I realised that I would "trash" a lot of work if my internet connection went down and nobody was around to get it going again. So I reduced my "connect to network" interval back to 0.2, did a manual update to get the change registered and went off happy in the knowledge that approximately every 5 hours, my box would contact the servers to report any previously uploaded results so that if things went down I would at least get some of the credit.
Well, Murphy's Law being what it is, you can imagine what happened. My box happily crunches away on those 10 units and uploads results as they complete. However, no contact was made with the scheduler. Why, you ask? Because my box didn't need to make contact as it had plenty of work on hand!! The box was down to the second last work unit and was within probably an hour or three of needing to contact the scheduler. There were eight uploaded but not reported results sitting in my queue. Then there was a tiny power flicker. How do I know this? Because of about 15 boxes running, about half went down (including the one running E@H) and about half stayed up.
When I got back and restarted the crashed boxes, the eight results were reported but because they had all expired in the meantime, all but one were granted no credit. It was just bad luck - very unfortunate timing. My mistake really was not understanding the full consequences of a drastic change in the "connect" interval from 2.0 days to 0.2 days. Since that time I've stayed on 0.2 days and increased the number of participating boxes dramatically. I've had no further problems with losing credit for results completed.
> I thought you can quote chapter and verse asap ;)
Not this either ...
When I got above 500 pages I started to have problems finding things ... I know it is in there SOMEPLACE, I just can't remember where ... So, I use the search many times ...
May I join this flame thread
)
May I join this flame thread ;)
My ***ing beta works just let me lose a WUs credit by less then 2h :(
Received 19 Apr 2005 16:08:41 UTC
Report deadline 19 Apr 2005 14:23:40 UTC
Should stick to my preferences, no fumbling in between WUs ;)
Grüße vom Sänger
> I missed it again. > My
)
> I missed it again.
> My result was uploaded toonight within the deadline but when i wake up in the
> morning reported too late only a few minutes.
>
I have a question along these lines. I understand that uploading and reporting is a two-step process... but what i don't understand is why it is a two-step process. Is there no way that results could be reported at the same time they are uploaded?
It seems quite a shame that Mike completed the workunit on time, but because the reporting process is delayed, he lost his credit.
Just a question, not a complaint. :)
Dig
> I have a question along
)
> I have a question along these lines. I understand that uploading and
> reporting is a two-step process... but what i don't understand is why
> it is a two-step process. Is there no way that results could be reported at
> the same time they are uploaded?
>
> It seems quite a shame that Mike completed the workunit on time, but because
> the reporting process is delayed, he lost his credit.
>
> Just a question, not a complaint. :)
I thought I had explained that ... but could only find mention that it is a two step process.
But the point is ... Ha! Found [url=http://boinc-doc.net/_phpMyFAQ/index.php?action=artikel&cat=291804&id=514&artlang=en]it!
[edit] :) I also define it
)
[edit] :)
I also define it as Reporting Process.
> [edit] :) Hey Paul, you
)
> [edit] :)
Hey Paul, you don't know your manual by heart?
I thought you can quote chapter and verse asap ;)
Grüße vom Sänger
> I thought I had explained
)
> I thought I had explained that ... but could only find mention that it is a two step process.
>
> But the point is ... Ha! Found [url=http://boinc-doc.net/_phpMyFAQ/index.php?action=artikel&cat=291804&id=514&artlang=en]it!
>
Paul, thanks for the link. I knew from reading the boards that the two-step process was necessary, i just never knew why it was necessary. ;)
Dig
> I have a question along
)
> I have a question along these lines. I understand that uploading and
> reporting is a two-step process... but what i don't understand is why
> it is a two-step process. Is there no way that results could be reported at
> the same time they are uploaded?
Why? - Because it saves the overhead of server contact. For immediate reporting:-
1. Upload result to data server
2. Report claim for credit to scheduling server
and sometime later...
3. Request more work from scheduling server.
For delayed reporting we only need:-
1. Upload result to data server
and sometime later...
2. Report claim for credit and request more work from scheduler.
> It seems quite a shame that Mike completed the workunit on time, but because
> the reporting process is delayed, he lost his credit.
Yes, it is a great shame. When I first started with one machine, I thought it would be a good idea (Seti classic mentality) to keep a goodly cache of work on hand, just in case.... I was about to go away on a one week business trip so I thought I'd "stock up". I had about 10 units on hand. Just before heading off to catch the plane, I changed my mind when I realised that I would "trash" a lot of work if my internet connection went down and nobody was around to get it going again. So I reduced my "connect to network" interval back to 0.2, did a manual update to get the change registered and went off happy in the knowledge that approximately every 5 hours, my box would contact the servers to report any previously uploaded results so that if things went down I would at least get some of the credit.
Well, Murphy's Law being what it is, you can imagine what happened. My box happily crunches away on those 10 units and uploads results as they complete. However, no contact was made with the scheduler. Why, you ask? Because my box didn't need to make contact as it had plenty of work on hand!! The box was down to the second last work unit and was within probably an hour or three of needing to contact the scheduler. There were eight uploaded but not reported results sitting in my queue. Then there was a tiny power flicker. How do I know this? Because of about 15 boxes running, about half went down (including the one running E@H) and about half stayed up.
When I got back and restarted the crashed boxes, the eight results were reported but because they had all expired in the meantime, all but one were granted no credit. It was just bad luck - very unfortunate timing. My mistake really was not understanding the full consequences of a drastic change in the "connect" interval from 2.0 days to 0.2 days. Since that time I've stayed on 0.2 days and increased the number of participating boxes dramatically. I've had no further problems with losing credit for results completed.
Cheers,
Gary.
> Hey Paul, you don't know
)
> Hey Paul, you don't know your manual by heart?
No ...
> I thought you can quote chapter and verse asap ;)
Not this either ...
When I got above 500 pages I started to have problems finding things ... I know it is in there SOMEPLACE, I just can't remember where ... So, I use the search many times ...