Can this please be explained?
I do see the time is posted correctly. After finishing the first 4 units, I did get the time the 4 units were worth. The Pending Time is very deceiving though. It's hard to judge when you will hit the next milestone. I guess I never noticed it before because I was doing a lot of short work, but the current work I am doing takes between 1.25 and 4 days depending on the processor. Right now there is over 2 days of processing on 16 cores, but the pending only shows enough for 4 cores. This would be nice if it was fixed.
/ 
Reply
Quote
I noticed there is a bug in Pending hours.
I know that noone of my hosts now crunching, for example, PrimeGrid 321 LLR, but there reporting 11.42 Pending hours in PrimeGrid 321 LLR.
The same thing with PrimeGrid Genefer.
Would you please explain or fix this bug?
/ Reply
Quote
I noticed there is a bug in Pending hours.
I know that noone of my hosts now crunching, for example, PrimeGrid 321 LLR, but there reporting 11.42 Pending hours in PrimeGrid 321 LLR.
The same thing with PrimeGrid Genefer.
Would you please explain or fix this bug?
If you were processing that type of work unit and it then finished and uploaded, the pending wont change under WUProp until the WUProp work unit finishes and uploads itself. It then tells WUProp project servers that the work unit has finished and then the pending updates.
Things only change after each WUProp work unit uploads and passes its data to the Project Server.
You may have to wait up to a maximum of 3 hours (the WUProp WU run time), depending when the a projects work unit uploads.
Conan
____________
/ Reply
Quote
You may have to wait up to a maximum of 3 hours (the WUProp WU run time), depending when the a projects work unit uploads.
Noone of my hosts has already crunched PrimeGrid 321 LLR in the last month.
And noone of my hosts has paused tasks of PrimeGrid 321 LLR.
What will your answer after that?
/ Reply
Quote
You may have to wait up to a maximum of 3 hours (the WUProp WU run time), depending when the a projects work unit uploads.
Noone of my hosts has already crunched PrimeGrid 321 LLR in the last month.
And noone of my hosts has paused tasks of PrimeGrid 321 LLR.
What will your answer after that?
That being the case, I suppose you could try and click 'update' for Primegrid (does not matter if it is set to No New Work), and see if after the WUProp work unit reports things have cleared up.
Otherwise don't worry about it too much as it is only a guide and is often off the mark by quite a bit.
As long as you get the correct hours recorded after the work units report then anything showing in pending does not really matter.
So far I have not noticed the problem that you are reporting but I will keep an eye open to see if it happens to me also.
Hopefully a Project Administrator will pop by the forums and look into the issue for you as I am out of ideas.
P.S. (Only you can see your computers so I did not know you had no Primegrid work on you computer, my suggestions were of a general nature.)
Conan
____________
/ Reply
Quote
Albert App Typo!
Albert@Home Gravitational Wave S6 Directed Serach (CasA) 94.20 0.00
...
Albert@Home Gravitational Wave S6 Directed Search (CasA) 0.00 12.28
Can they be merged?
____________
=sticky tape=
/ Reply
Quote
How can I prove my words?
For example in this way.
My WUProp stats:
My PrimeGrid tasks:
As you can see there are no PrimeGrid 321 LLR tasks in progress.
(there only 1 task completed and waiting for validation, but for WUProp this task has already counted to Running hours as far as I understand)
The same thing with PrimeGrid Genefer.
Anyone can make sure, my PrimeGrid account is opened for viewing:
http://www.primegrid.com/hosts_user.php?userid=64131
/ Reply
Quote
The problem I see with some of those it seems that if WUProp site/database is down for a while (backup/clean up, ISP issues, etc) and you have finished one, but cannot get a new one for a little bit, then WUs finish, nothing is running to see them finish. Then when WIProp starts up again it gets confused and just let's the pending hours sit because it is unsure what to do with the information.
/ Reply
Quote
Hal has been down for ages and the Pending just sits there, FreeHAL@home FreeHAL 2012 app 5,371.57 502.40
Not that it matters after reaching Blue!
____________
=sticky tape=
/ Reply
Quote
I have noticed Pending Hours are much more in line on the CPUs now. So if I am running 12 instances of a projected on several machines, the pending and what has passed is very close.
Thanks.
/ Reply
Quote
I've been running 3 instances of SLinCA for 38h each, but pending hours is only reporting one x 38h (up to the last WUProp upload),
SLinCA@Home heavy_lodci011eq32v0p125T1XZ40 - MIN:runtime~32h(checkpoint 2h),RAM~1GB,HDD~60MB 0.00 37.47
It should be 114h.
Linux - Ubuntu 13.04 (x64).
/ Reply
Quote
Pending hours do not expire when a unit expires and/or is purged. I had a computer fail and I am not planning on bringing it back online. In fact the drive has already been wiped and the unit is being donated off.
So, will there pending hours sit forever?
/ Reply
Quote
Pending hours do not expire when a unit expires and/or is purged. I had a computer fail and I am not planning on bringing it back online. In fact the drive has already been wiped and the unit is being donated off.
So, will there pending hours sit forever?
Since the hours are still sitting there, let me expand on this.
I had a computer go belly up. It was crunching Sophie Germain and PPS Sieve at the time. I am unsure how many hours the PPS Sieve are sitting there because I am working on it with other computers, but Sophie Germain has 2,72 pending hours that are dead. The WUs will never be returned as I have already wiped the machine and donated it off to a good cause.
So will this time always sit there as Pending? The computer in question has already been deleted from my account. I am unsure how easy it is to remedy the situation of a computer dying/stops sending back work.
/ Reply
Quote
OK, I see another anomalously. When doing multiple projects on a single machine, the Pending hours do not calculate correctly. For example I have 4 cores running 2 each of 2 different projects. The longer term project has been running 15+ hours per core, but the pending only shows 15+ hours instead of 30+. Short term ones are the same way. Only counting one core worth of time. If I run all 4 cores on the same project then the Pening Hours is much better.
/ Reply
Quote
Nobody cares...
To count Pending Hours correct they need to keep figures for every host and
- calculate total Pending Hours by all hosts after every user WU report
- update host's Pending Hours after every host WU report
- delete host's Pending Hours or decrease total Pending Hours after host expired
But it's difficult... so nobody cares.
/ Reply
Quote
Is there a way to clear pending hours that did not clear from a machine dying? I have two projects that currently have dead time. These should auto clear of a machine times out the next WU and it's purged.
/ Reply
Quote
I have a different issue, all my pending SLinCA work has been cleared to zero even though the work units are still on my computers due to the server being down over a week, the computers are still working just can't upload that project.
Yet some work that is still on a different computer that crashed is still showing its pending hours even though that computer has been down for about 3 weeks.
Conan
____________
/ Reply
Quote
I have 1300+ hours pending in Suduku according to my pending hours list here. However, I haven't had any systems really doing much over there for a few months since I hit 1 million points. I doubt I have that many unreported WUProp work units out there, so something isn't quite working.
____________
/ Reply
Quote
[AF>WildWildWest]Sebastien
Dictator
Send message
Joined: 28 Mar 10
Posts: 2424
Credit: 493,569
RAC: 52
Total hours: 1,291,388
The host 39150 run sudoku workunit
____________
/ Reply
Quote
http://sudoku.nctu.edu.tw/show_host_detail.php?hostid=14020
Not 1300+ hours it isn't. That device has not been active at that project since April. I only have One device that checks in there and it isn't that host. 2 virtual machines that are not accepting work. Those are the only active devices for a few months. (June 26th was the last device before those 3) So, I think you are mistaken.
http://sudoku.nctu.edu.tw/hosts_user.php?sort=rpc_time&rev=0&show_all=0&userid=242
____________
/ Reply
Quote