| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Official Messages
Forum: News
Thread: 2022-09-13 (Network & Contributions Page Mini-Update) |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 65
|
|
| Author |
|
|
TPCBF
Master Cruncher USA Joined: Jan 2, 2011 Post Count: 1932 Status: Offline Project Badges: 
|
So, I do not trust the results statistics at all for the moment, and I seriously doubt that they will be able to backfill the result lists, from the date when the test period at Krembil begun. If they're not able to do that, at least they must synchronize the current BOINC credits people have at WCG, with the WCG points. I do have a hunch for weeks now that there is something wrong with the stats numbers, even before the WCG stats (for current returns) started working. Due to time constraints and the lack of cooperation of the WCG web site, I have not been able to clearly pinpoint this. I think that with one particular project, there is a mixup of the difference between WCG points and BOINC points and a "adjustment" is made the wrong way, or one time to many... Getting busier at work right now, as well as having to deal with two cracked ribs at the same time, I am not sure when I will have time to try and track this down (or possibly, refute my hunch). This was the main reason why I started to look into creating my own stats program based on the history.csv file from the Result page... But in any way, in order to do this more conclusively, it would require that the web site and downloads are working for some time reliably. Which simply has not been the case, the downloads are still screwed up for the 6th day in a row now...  Ralf  |
||
|
|
erich56
Senior Cruncher Austria Joined: Feb 24, 2007 Post Count: 294 Status: Offline Project Badges: 
|
... and I seriously doubt that they will be able to backfill the result lists, from the date when the test period at Krembil begun. that's exactly what I doubt, too as it looks now, they are not even able to backfill missing results from 3 days ago: for October 4th, under "History" the figure of 24 returned results is still shown, whereas the correct figure would be about 1.500 ! |
||
|
|
wildhagen
Veteran Cruncher The Netherlands Joined: Jun 5, 2009 Post Count: 762 Status: Offline Project Badges: 
|
... and I seriously doubt that they will be able to backfill the result lists, from the date when the test period at Krembil begun. that's exactly what I doubt, too as it looks now, they are not even able to backfill missing results from 3 days ago: for October 4th, under "History" the figure of 24 returned results is still shown, whereas the correct figure would be about 1.500 ! True. The stats run of this morning shows I have completed 13 work units. In reality this should be something between 500 and 600 for a half day. Spread over 120+ cpu cores plus 2 GPUs. Apparently the stats are as unreliable as Krembil itself, the figures do not reflect the reality, not even close. |
||
|
|
TPCBF
Master Cruncher USA Joined: Jan 2, 2011 Post Count: 1932 Status: Offline Project Badges:
|
True. The stats run of this morning shows I have completed 13 work units. In reality this should be something between 500 and 600 for a half day. Spread over 120+ cpu cores plus 2 GPUs. Most everyone will agree that WCG under Krembil isn't quite the pillar of reliability, but are you guys sure that there even was a stats run this morning? Apparently the stats are as unreliable as Krembil itself, the figures do not reflect the reality, not even close.  Ralf |
||
|
|
wildhagen
Veteran Cruncher The Netherlands Joined: Jun 5, 2009 Post Count: 762 Status: Offline Project Badges:
|
True. The stats run of this morning shows I have completed 13 work units. In reality this should be something between 500 and 600 for a half day. Spread over 120+ cpu cores plus 2 GPUs. Most everyone will agree that WCG under Krembil isn't quite the pillar of reliability, but are you guys sure that there even was a stats run this morning? Apparently the stats are as unreliable as Krembil itself, the figures do not reflect the reality, not even close. Ralf Must be, as the history page shows a entry for today, 7 October, with in my case 13 workunits, which is absolutely impossible. My primary work machine alone returned tens of work units, let alone my multiple other pcs and servers. |
||
|
|
TPCBF
Master Cruncher USA Joined: Jan 2, 2011 Post Count: 1932 Status: Offline Project Badges:
|
Must be, as the history page shows a entry for today, 7 October, with in my case 13 workunits, which is absolutely impossible. My primary work machine alone returned tens of work units, let alone my multiple other pcs and servers. If with "history" you are referring to the csv file under "My Activity over time", yes, that is off (I certainly returned more than 16 MCM1 and 2 OPN1 WUs in that time frame), but with that half day outage of the web site and the forum, I would not expect this is include anything transferred that would fall into that time frame. But those WUs do seem to show up in the history,csv file under "Results", so there is a strong likelyhood that these will show up after the next run. Restoring WU stats from June through 9/28, that certainly won't happen any time soon and then likely piecemeal...Ralf |
||
|
|
phillipspencer
Advanced Cruncher France Joined: Apr 9, 2015 Post Count: 71 Status: Offline Project Badges:
|
True. The stats run of this morning shows I have completed 13 work units. In reality this should be something between 500 and 600 for a half day. Spread over 120+ cpu cores plus 2 GPUs. Most everyone will agree that WCG under Krembil isn't quite the pillar of reliability, but are you guys sure that there even was a stats run this morning? Apparently the stats are as unreliable as Krembil itself, the figures do not reflect the reality, not even close. Ralf Must be, as the history page shows a entry for today, 7 October, with in my case 13 workunits, which is absolutely impossible. My primary work machine alone returned tens of work units, let alone my multiple other pcs and servers. The stats for today do look pretty weird. I have just 1 MCM and nothing (not even a zero) for OPN and I know yesterday many more WUs were returned. Overall, since stats restarted on 29 September I had been thinking the hours added looked very low and this seems to confirm it. Hopefully, it is just stats lost and not the WU science. |
||
|
|
TPCBF
Master Cruncher USA Joined: Jan 2, 2011 Post Count: 1932 Status: Offline Project Badges:
|
[Overall, since stats restarted on 29 September I had been thinking the hours added looked very low and this seems to confirm it. I don't think so.There are a couple of things to consider here. First of all, the first stats run on 9/29 didn't catch all the WUs of that "normal" time frame, and the download issue where likely also still partially present during that time. For the next couple of days, there were no download issues, and for me, the shown stats showed what seems to be a matching representation of the work returned, on those days. Then the download issues came back, with several hosts that couldn't be babysitted running out of work. That also resulted in a drop of WUs in the stats. And while the web site is off, this also seems to effect the process that run the stats. So yes, for all but Wednesday/Thursday last week, the stats will show lower numbers of WUs returned... Ralf |
||
|
|
erich56
Senior Cruncher Austria Joined: Feb 24, 2007 Post Count: 294 Status: Offline Project Badges:
|
in the past few hours, uploads get stuck very often.
So now there is not only the problem with the well-known network problems affecting the downloads, but uploads, too. So, all in all, the situation is getting even worse, instead getting better |
||
|
|
TPCBF
Master Cruncher USA Joined: Jan 2, 2011 Post Count: 1932 Status: Offline Project Badges:
|
in the past few hours, uploads get stuck very often. Yeah, Krembil sure likes to mix things up, so we never get bored. Always something new to get frustrated about... So now there is not only the problem with the well-known network problems affecting the downloads, but uploads, too. So, all in all, the situation is getting even worse, instead getting better  Ralf |
||
|
|
|