[b]YAY! BACK TO NORMAL!

---

> When they will fix it?
>
Soon ;)

When they will fix it?

---

> Paul, you're obviously one of the accounts not affected.

I just looked again an my numbers changed!

Ok, so I don't know what I am talking about. :)

If it was easy, anyone could play ...

---

> Just as an eample, I got my total credit for each project from my account and
> compared it to the numbers of the stat site:
>
> LHC 29,452 vs. 29,452
> CPDN 212,021 vs. 211,643
> EAH 31,726 vs. 31,014
> PAH 56,523 vs. 56,257
> SAH 171,468 vs. 169,137
>
> So, as you can see, there is a lag built in and the stat site has me lower by
> a fair margin. But since the numbers all update at the same time, you can
> keep track of where you are reletive to everyone else ...
>

Paul, you're obviously one of the accounts not affected.

I understand all about lag, both in Berkeley producing the stats, the doenload and update.

I don't know which stats site you use, but at BOINCSynergy (http://www.boincsynergy.com/stats/boinc-stats.php?id=3336935&project=sah) It shows the past 8 days credit.

In my case:

06 Apr: 587
07 Apr: 592
08 Apr: 472
09 Apr: 166
10 Apr: 75
11 Apr: 186
12 Apr: 29

The 6th and 7th are the figures I expect. The 9th, 10th, 11th and 12th are plain wrong.

My previous weeks credit was also in excess of 500 per day, and I still have the same 3 systems crunching.

I could understand it if 1 or even possibly 2 days were skewed by delays/failures, but the decline in reported credit started 5 days ago on the 8th. My real-time credit is also higher than the site credit by about 2500 = the missing few hundred on each of the 8th/9th/10th/11th/12th.

I have also downloaded the user.xml (user.gz) myself several times and the figures stored therein agree with the figures on the site(s).

The last user.xml I downloaded is dated the 12th and still does not reflect the real-time credit at that date or any of the other short-credit dates.

Something is wrong at Berkeley - the stats extract is broken.

---

Just as an eample, I got my total credit for each project from my account and compared it to the numbers of the stat site:

LHC 29,452 vs. 29,452
CPDN 212,021 vs. 211,643
EAH 31,726 vs. 31,014
PAH 56,523 vs. 56,257
SAH 171,468 vs. 169,137

So, as you can see, there is a lag built in and the stat site has me lower by a fair margin. But since the numbers all update at the same time, you can keep track of where you are reletive to everyone else ...

---

I have the exact same conclusion, that the users.xml is late with the updating. Because it just started to do that, will be a drop in stats as cretits/day for a while. When the users.xml will come back closer in time, the stats will inflate artificially (kinda of Doppler effect since is a frecqvency (of units per day) and a speed of "detecting" those pulses ( WU)into users.xml (the receiver)

---

> Hello,
> I noticed too that difference; until 4 or 5 days ago all worked fine but now
> it doesn't:
> In the early hours of April 11th I uploaded about 500 wus; Boinc client says
> my credit before uploading was 222,400; after uploading 243,140;
> the boincstats does not say that I earn about 20,000 credits, it says I eran
> 30 or 40 credits per day and that I have now 223,900 credits (in boincstats).
> Is maybe the www.boincstats.com site broken?
> Regads
> Roberto
>

No, it is not the stats site(s) that is/are broken.

a) I see the same problem on two different stats sites.

b) I downloaded the users.xml myself and can see the figures in the stats file are those shown on the sites. My entry in users.xml does _not_ agree with 'My Account' page.

c) The stat's file is dated 12-Apr-05 11:13 and contains "Total Credit"="My account" about 48-72 hrs ago.

---

Hello,
I noticed too that difference; until 4 or 5 days ago all worked fine but now it doesn't:
In the early hours of April 11th I uploaded about 500 wus; Boinc client says my credit before uploading was 222,400; after uploading 243,140;
the boincstats does not say that I earn about 20,000 credits, it says I eran 30 or 40 credits per day and that I have now 223,900 credits (in boincstats).
Is maybe the www.boincstats.com site broken?
Regads
Roberto

---

Hmmm... Perhaps some developer is fiddling with the stats file format? Remember it was also invalid a short time ago?

---

Hi

What i can see is that my credits increases constantly and quick, but the stats site didnt update.

I´m 2 days at 172xxx but in stats i´m at 170xxx.

Maybe a problem with exported xml stats.

greetz Mike

---

With each crime and every kindness we birth our future.

> Is it possible that Seti stopped other sites from accessing the STATS until
> it's done with the database Purging? This would mean it's stats are correct,
> but exported ones aren't. just a thought.

If that is the case, they didn't do a very good job at it : the stat sites still seem to increase their seti stats, though at a fraction of the actual increase.
So I think that is not the case

---

Is it possible that Seti stopped other sites from accessing the STATS until it's done with the database Purging? This would mean it's stats are correct, but exported ones aren't. just a thought.

tony
for example, if it's indexing the drive in purging the data than maybe they can't get in.

> So, Credit seems slow the last 3 or so days? Hmm, The file Deleter/purger has
> been working the last 3 or so days. I wonder if it's a coincidence?
>
> tony
>

The stats on "Your account" seem to increase as should be expected.

The problem lies with the stats that are exported : different stat-interpreting sites (www.boincstats.com and boinc.mundayweb.com) have the same results : for other projects, the stats seem to be ok, but for seti, about 1/3rd of the new credits seems to be all that gets added.

I haven't checked with too many accounts, but all accounts that I checked, displayed the same behaviour : the difference between the stats on the statsites and the stats on "Your account" seems to get bigger by the day.

---

Ok, so if v4.25 grants less credit because of the revised benchmarking
code, then I would have been effected by this even if I didn't upgrade
because SETI takes the middle value of the claimed credit in the quorum.
It is interesting though that the top crunchers were not as effected by
this as most of us are, as my total RAC used to be an average of around
6000-6100, now it is 3677.22, and yet there are still users with over
14000 RAC, and are still in the 6000+ range down to 21st place in RAC.
I have been continuously replacing slower computers with much faster
ones (I have 17 machines total), and my RAC rank has been dropping off
the map. I was at one point ranked at 41st place for RAC about 4 upgrades
ago, and now I am down to 68th. I figured I should at least be able to
hold my place, if not gain some ground by replacing 4 older machines
with state of the art machines, but it all seems to have the opposite
effect. I guess I would just like to know what I can do to reverse this
depressing trend. The faster my machines get, the more my RAC drops...
It just doesn't seem logical...

Regards, Daniel.

---

So, Credit seems slow the last 3 or so days? Hmm, The file Deleter/purger has been working the last 3 or so days. I wonder if it's a coincidence?

tony

How do the stats sites compare with "Your Account" page here?

I ask because I've noticed the same lack of credit on stats sites. Over the past 72 hours, 'My Account' has show total credit rising from around 80,700 to 82,500, but the stats sites I've looked at have only show my daily credit as between 100-200 CS, giving a under-stated total of only about 81,200. There is well over 1000 CS missing somewhere.

I happen to believe that the 'My Account' figure is accurate. So the Validator seems to be OK.

As several stats sites don't agree in the same manner, I think the stats export at Berkeley must be a tad broken.

---

> But I can say definitively that the rapid decay of credit
> over the last 2-3 weeks is NOT due to a version change, as I just changed
> versions 2 days ago.

But don't forget it was 3 or so weeks ago that V4.25 was released, the one with the revised benchmarking code.
You don't necessarily get the credit you claim, just what's left after everyone elses claims have been considered...

---

Grant
Darwin NT

I too have noticed a dropoff in credit. I just pulled an Opteron 244 CPU out
of one of my machines and replaced the single CPU with a PAIR of Opteron 246's a couple of weeks ago. Since I did that, I seem to be producing a lot LESS credit. And it seems to have affected more than one machine, as evidenced by the RAC numbers from my machines. My top cruncher, a dual Opteron 248 system, has been consistantly producing RAC numbers of between 799 to 845. Now it is down to 655.67 and falling daily. The numbers for what used to be the single Opteron 244 (now dual 246's) have gone up a bit, but all the rest of my
machines RAC is decaying rapidly. 2 days ago, I saw that BOINC version 4.25
is out, so I updated from version 4.19 on the dual Opteron 248 and my FX53
system. But so far, I am unable to tell any difference, although it is really
too soon to tell. But I can say definitively that the rapid decay of credit
over the last 2-3 weeks is NOT due to a version change, as I just changed
versions 2 days ago. I am getting my numbers directly from my account page
here on the SETI web page.

Regards, Daniel.

---

I also see that sites like boincstats & mundayweb give but about 30% from the credit that's actually gained, as can be seen on my seti profile page. This has been going on for at least 2-3 days.

Could it be a problem with the exported xml file?

---