Pages: [1]
Guest

2005-04-10 08:59:21

Hi, as far as I can see the stats for the Team Anadtech are not correct:

Compare please
[url=http://www.boincstats.com/stats/team_graph.php?pr=einstein&id=12' target='_blank]http://www.boincstats.com/stats/team_graph...=einstein&id=12[/url]
with
[url=http://macg.no-ip.info:5520/boinc/teams.php?proj=einstein' target='_blank]http://macg.no-ip.info:5520/boinc/teams.ph...p?proj=einstein[/url] and
[url=http://www.boincsynergy.com/stats/teams.php?team=12&project=eah' target='_blank]http://www.boincsynergy.com/stats/teams.ph...=12&project=eah[/url]

When we add up the daily production for inidividual members the totals presented do not add up: it takes only 4 team members to produce more daily creds than on the page presented. I have seen that this is not correct for at least 2, if not 3 days.
BTW the magc and boincsynergy-pages give the same numbers ...

Could this be a problem for other teams than team Anandtech?
Guest

2005-04-10 09:43:16

Hi,
This may be the same problem as the one I described above:
When I check my personal stats for Einsten@home, the totals do not add upp either. According to some other sites [url=http://macg.no-ip.info:5520/boinc/team.php?proj=einstein&teamid=12' target='_blank]http://macg.no-ip.info:5520/boinc/team.php...stein&teamid=12[/url]
I have a total credit of 38863 for april 9.
looking at the top of [url=http://www.boincstats.com/stats/user_graph.php?pr=einstein&id=20949' target='_blank]http://www.boincstats.com/stats/user_graph...nstein&id=20949[/url]
I have the same 38862,89 creds but in the table of "last 30 days" the creds for april 9 totals to 36,663. If this error "goes on" to the team statistics, then there may be some bug.
[BOINCstats] Willy
 
Forum moderator - Administrator - Developer - Tester - Translator
BAM!ID: 1
Joined: 2006-01-09
Posts: 9461
Credits: 353,172,950
World-rank: 5,070

2005-04-10 10:04:16

Regarding your reported bug #1: when I add up the credit-a-day from all Anandtech members for Einstein i get 4203. This is equal to the credit-a-day from the team. I see no problem here.

As for your bug #2: please READ. The stat on the bottom is based on the last daily update. The credit on top is from the last incremental update. This is also reported here: [url=http://www.boincstats.com/forum/viewtopic.php?t=170' target='_blank]http://www.boincstats.com/forum/viewtopic.php?t=170[/url].
Guest

2005-04-10 10:40:53

Thanks Willy, for the fast response.
So, the problem is somewhere else: because according to the stats from the Einstein@home - homepage the 5 top crunchers in Team Anadtech produced april 9 together 7527 credits, which is more than 4203. I myself produced 2200 credits according to - amongst others - [url=http://www.boincsynergy.com/stats/teams.php?team=12&project=eah' target='_blank]http://www.boincsynergy.com/stats/teams.ph...=12&project=eah[/url]. So: Why do the team numbers from e.g. [url=http://einstein.phys.uwm.edu/top_teams.php' target='_blank]http://einstein.phys.uwm.edu/top_teams.php[/url] differ so much from boincstats?
:roll:
Saenger
 
Tester - Translator
BAM!ID: 5
Joined: 2006-01-10
Posts: 1735
Credits: 231,802,342
World-rank: 6,771

2005-04-10 11:38:32

Hello Petrusbroder,
where is the huge difference between:

BoincSynergy:
Total credit granted: 706,288 cobblestones
Total Recent Avg Credit: 14,439 cobblestones
Team Rank TC: 13 of 2,014
Team Rank RAC: 14 of 2,014

and BoincStats:
Total Credit 706,288.88
World Position 13
Recent Average Credit 14,446.33398

And if you look at your teams credit per day here on this site
( [url=http://www.boincstats.com/stats/graph2.php?pr=einstein&table=teams&id=12' target='_blank]http://www.boincstats.com/stats/graph2.php...ble=teams&id=12[/url] ), it droped really far for the last day. So my suggestion would be, that the XML from the project were not correct yesterday, or updated at a wrong time.

BTW, same happened with my team, so nothing to worry about:
[url=http://217.67.229.234/stats/graph2.php?pr=einstein&table=teams&id=49' target='_blank]http://217.67.229.234/stats/graph2.php?pr=...ble=teams&id=49[/url]

Patience is a virtue
Grüße vom Sänger
Guest

2005-04-10 12:33:40

Well, I do not know about XLM-files and where those may go wrong. And I am not blaming anybody at all, I am just trying to get som consistency from the diffrent stat-sites.

So if I check this site
[url=http://www.boincstats.com/stats/team_graph.php?pr=einstein&id=12' target='_blank]http://www.boincstats.com/stats/team_graph...=einstein&id=12[/url]
the last 30 days, I see the following:
2005-04-06: 650902 credits and 17165 credits/day
2005-04-07: 667667 credits and 16776 credits/day
2005-04-08: 683034 credits and 15357 credits/day
2005-04-09: 687237 credits and 4203 credits/day (pls see below!!!)

if i check this site:
[url=http://macg.no-ip.info:5520/boinc/team_history.php?teamid=12&proj=einstein' target='_blank]http://macg.no-ip.info:5520/boinc/team_his...2&proj=einstein[/url]
the last 4 weeks, I see the following:
April, 6: 650902 credits, 17165 credits/day (<--- perfect)
April, 7: 667667 credits, 16776 credits/day (<--- perfect)
April, 8: 687237 credits, 19560 credits/day (difference compared to boincstats: 4203 credits)
April, 9: 706289 credits, 19051 credits/day (difference compared to boincstats: 19052 credits)

To me a differenc of 19052 credits is huge, since until now all the diffrences I have seen are in the order of magnitude 10 - 100 credits.
It is also interesting to note, that the difference between the two sites on april 8 equals the daily production on april 9. Hmm. This kind of coincidence should not happen by chance.

So: by looking at the numbers, it is clear to me, that in boinc stats there is an misalignment (?) of the dates. Well I did not analyse the numbers this closely when I posted my first post, so now I know that there may be an error in the imported file. But: how can I know werther this is a bug in the script generating the XML-file or that this is a bug in the script which extracts the data from the XML-file. That is not translucent to me as a user. Since I see, that the stats from macg.no-ip.info fit my own stats and that those from boincstats do not, of course I post to the site where the difference is.

Also, since I have been working in statistics in the 1980-ies, I do know, that small errors accumulate. I do not know, whether these differences arise from XLM-files you obtained or from somewhere else. I thought that I could be helpful and make you aware of a problem.

Really no offence intended. I admire you guys who do this job in yor free time. Bu there is a chance that there is a bug (AFAIK) and I thought that that may be interesting for you.

I may not be so clear in my presentation of the problem, but, not being a programmer, and not knowing the exact language and the exact words for differenet things in english or german or french, I use the language I think may work. I certainly did not / do not want to sound bossy or impertinent and If I do/did it was not intentional and I apologize.

Since I am a very courious person, I still would like to understand what goes on, so if you have an explanation, please let me know.
Guest

2005-04-10 21:39:59

I'd like to confirm the observation of Petrusbroder - and I'm quite sure, that Willy has noticed this problem very well.
However, to me it looks like all teams have the same 'problem': an additional readout of daily production in late 9.4 or early 10.4 - and this is very low.
No wonder, since it was only a quarter of a day.
In the long run, this will not matter, but at first glance it looks like un ugly bad performance in early april05.

But since everyone has it...

It may also be fixed, though.
:wink:
[BOINCstats] Willy
 
Forum moderator - Administrator - Developer - Tester - Translator
BAM!ID: 1
Joined: 2006-01-09
Posts: 9461
Credits: 353,172,950
World-rank: 5,070

2005-04-11 05:02:49

I see that all teams have less credit for that day. But this is only in Einstein, not the other projects. As the update is an automated process and it works with XML files from the projects, I have little influence on the outcome of the stats. There was no extra daily update.

The other stat sites probably update at another time then BOINCstats. Maybe Einstein exported their XML file after BOINCstats updated. Or it exported early the day before. Either way, the numbers are correct when compared to the other sites.
Guest

2005-04-11 18:36:42

Thanks Willy, for your efforts and answer. I appreciate it.
Best regards from a still -partly- snow covered northern Sweden, where the reindeers still frolic in the snow only 30 km from any big city.... :wink:
Guest

2005-04-12 10:10:46

Thanks for the hard work on a great site! There does seem to be a definite problem with Seti however. My RAC runs around 1000/day but over the last few (4) days only 300 or so are being reported. At the moment the total credit is off by about 3000. I don't watch CPDN as closely so I'm not sure if that one is reporting the same way.
Guest

2005-04-12 12:34:10

There is indeed a problem with the seti stats, but I don't think that the problem comes from boincstats.com. Boincstats is not the only statsite that has these problems with seti stats.

As far as the accounts that I have checked, the credits on "Your account" at seti seem to be correct, but not in the exported stats (didn't check the data in the actual xml files though).
Guest

2005-04-12 19:55:04

There is indeed a problem with the seti stats, but I don't think that the problem comes from boincstats.com. Boincstats is not the only statsite that has these problems with seti stats.

As far as the accounts that I have checked, the credits on "Your account" at seti seem to be correct, but not in the exported stats (didn't check the data in the actual xml files though).


I'd like to confirm that observation. I just tried boincsynergy and they have the same numbers for my seti account. The problem is that they have been the same or slight change for three straight days even though on the setiweb.ssl.berkeley site I can clearly see the numbers climbing. Currently the stat pages here & synergy say I have <1000 credit but it's been >1000 for several days (up to >1300 now). I don't know why the lag (well over a day) but it's there. Perhaps by the end of the month it'll show me over 1000.
[BOINCstats] Willy
 
Forum moderator - Administrator - Developer - Tester - Translator
BAM!ID: 1
Joined: 2006-01-09
Posts: 9461
Credits: 353,172,950
World-rank: 5,070

2005-04-12 20:16:41

Normally your credits on the stats sites will synchronize with the project every day at least once. SETI however always has problems with the huge user base is has. They are actually the only real test-BOINC-project (IMHO). They run into problems first, and they solve them (after a while). So maybe they hit a wall again.

If BOINCstats and Synergy are in sync you can be sure the problem is with the project.
Guest

2005-04-13 08:39:16

I have noticed that on my machines where machine details are "hidden" in the profile, results are not being updated, but on my machines where the details are not "hidden" in the profile, they are being updated.

It might therefore be the case that the query on the server at Seti that genrates the response might have been changed in the last little while to hide all details concerning the machine that are hidde rather than just the hardware details.
Saenger
 
Tester - Translator
BAM!ID: 5
Joined: 2006-01-10
Posts: 1735
Credits: 231,802,342
World-rank: 6,771

2005-04-14 21:34:35

Look [url=http://setiweb.ssl.berkeley.edu/forum_thread.php?id=13738#98823' target='_blank]this thread[/url] over @ seti for this explanation:

Hi,

Sorry about the lagging stats. Mea culpa etc.

A whille ago I mentioned that we had switched to the new Operton server and were using the old Sparc machine as replica backup.

We run a daily big stats collection and found that it impacted the new server causing some slow queries while it ran. So we decided to run the stats collection on the replica server, which is not a problem when it is up-to-date with the master.

However, it is currently lagging by about 4 days because of some heavy db rows cleanup that we did on the master. We expect that the replica will catchup once it gets through all the deletes (the slowest db process).

In the meantime we will point the stat collection at the master db server that may slow the server response for a couple of hours daily. However, there will be a disjoint in the stats report once the information starts being gathered off the master, so please expect that by sometime tomorrow.

Thanks,

Bob Bankay,
BOINC_SETI DBA
Grüße vom Sänger
Guest

2005-04-20 11:52:43

I've been seeing some oddities that don't seem like "db" issues but rather "math error" issues.

SETI = 2155
Predictor = 824
BOINC = 742

I can think of no straight-forward mathmatical operations which combine 2155 & 824 and arrive at 742. Clearly they are not summed as they should be. And no, Synergy is not reflecting the same error at this point.[/code]
[BOINCstats] Willy
 
Forum moderator - Administrator - Developer - Tester - Translator
BAM!ID: 1
Joined: 2006-01-09
Posts: 9461
Credits: 353,172,950
World-rank: 5,070

2005-04-20 17:18:02

I've been seeing some oddities that don't seem like "db" issues but rather "math error" issues.

SETI = 2155
Predictor = 824
BOINC = 742

I can think of no straight-forward mathmatical operations which combine 2155 & 824 and arrive at 742. Clearly they are not summed as they should be. And no, Synergy is not reflecting the same error at this point.[/code]


[url=http://www.boincstats.com/forum/viewtopic.php?t=200' target='_blank]Bug submit guidelines![/url]
[BOINCstats] Willy
 
Forum moderator - Administrator - Developer - Tester - Translator
BAM!ID: 1
Joined: 2006-01-09
Posts: 9461
Credits: 353,172,950
World-rank: 5,070

2005-04-20 18:02:49

I've been seeing some oddities that don't seem like "db" issues but rather "math error" issues.

SETI = 2155
Predictor = 824
BOINC = 742

I can think of no straight-forward mathmatical operations which combine 2155 & 824 and arrive at 742. Clearly they are not summed as they should be. And no, Synergy is not reflecting the same error at this point.[/code]


I think you refer to your own [url=http://www.boincstats.com/stats/boinc_user_graph.php?pr=bo&id=5cef4ee669b3ef16e97a0917835300a1' target='_blank]account[/url] (found the link for you). The numbers add up nicely.
Guest

2005-04-21 11:41:48

[url=http://www.boincstats.com/forum/viewtopic.php?t=200' target='_blank]Bug submit guidelines![/url]


Ouch.

OK. Here ya go. The original message told you what it was and what I think it ought to be, here's the link:

[url=http://www.boincstats.com/stats/banner.php?cpid=5b627a4a8d262e4f4a81844654cfc68d' target='_blank]My Stats[/url]

But I think I discovered the cause of my problem. I just renavigated to the link on the web page (not where I'd stored it) and found that it had changed slightly (probably when I'd done some account fiddling). Still odd in that it is partially updating, partially not.

Thanks for listening.
[BOINCstats] Willy
 
Forum moderator - Administrator - Developer - Tester - Translator
BAM!ID: 1
Joined: 2006-01-09
Posts: 9461
Credits: 353,172,950
World-rank: 5,070

2005-04-21 19:30:32

You are using an old link. Your CPID probably changed. The link you use will be invalid in a few days.

Try [url=http://www.boincstats.com/stats/banner.php?cpid=5cef4ee669b3ef16e97a0917835300a1' target='_blank]this[/url] one instead.
Pages: [1]

Index :: BOINCstats Bug Report :: Incorrect stats?
Reason: