Is it just me or are the Stats not updating?[Nov 12 problem]

[bruce]

The other thing I speculated about was lack of competence on the part of those responsible for re-crediting. Look at this donor's daily chart on Extreme Overclocking's stats. Now look at this one. Doesn't that seem odd? I'm just asking questions here.

That's a good question, but out of the scope of the Nov 12 server crash. Moreover, the PPD summary gives us no information about what clients are folding nor does it identify why some clients may have stopped folding.

If I look at Plazzman's statistics, he has 6 clients which have returned WUs since Nov 12. He has 13 clients which returned WUs since Oct 29 which have NOT returned any results since Nov 12. Why? We can speculate that he shut down some clients or we can speculate that there's something else wrong, but we simply do not have enough information to identify why he went from either 19 clients to 6 clients or from 13 clients to 6. (I can't tell which one.)

Plazzman
Active clients (within 50 days) 10
Active clients (within 7 days) 1
As far as I can tell, he has not asked for help on this forum. In any case, we can't even tell if there's a problem to be solved or just that he took some hardware off-line. [Someone with a heat-related problem shutting down for the Summer in the Southern Hemisphere???]

Wolf
Active clients (within 50 days) 9
Active clients (within 7 days) 5
I see the same sort of issue with Wolf . . . and he's here . . . but he hasn't given us any information that will help fix the problem. Since 9/9 he has had 14 clients return a WU. Since Nov12, he has had six. I have no way of knowing if some of those clients are re-installs and represent the same hardware or if some of them have been shut down or if some of them are unable to return results unless he explains it to us.

I don't understand what's happening with FAHReQ, but it's probably somebody who has re-enqueue WUs but they're not being credited to his name. Like the other examples, if you can offer an explanation of what clients he is SUPPOSED to be running it would be very helpful. Maybe these are the WUs which were re-credited by Stanford for situations where they had a team number but the user-name was undecipherable.

(There is a reason why Stanford says: "You may choose any name you like, as long as it contains only letters or numbers (to insert a space, you may use the underscore "_" character)." I suppose there are reasons why people intentionally use obscure characters in their name -- but if the re-credited script fails to retrieve their name from the backup, they cannot expect to receive credit for lost WUs.)

Wolf and/or Plazzman: If you're reading this, start a new thread describing your problem, if you have one, and we'll work with you to get all of your clients folding again. If you know what happened to your other clients, let us know that, too. (The same applies to the rest of you.)

[Grandpa_01]

frobozz is on Team ATI, and has 37 clients active in the last week. And half of them are GPUs, and all reporting in the last few days, well after the Recredit was so called done by Bruce. So it's not a glitch, not a mistake, not cheating, just hard folding. That kind of hardware would certainly earn those kinds of points.

Don't claim a problem just because you don't understand the circumstances behind the data.

I don't know what "incorrectly" means, but however they did it, it was done as best they could with the resources they had. No cheating. No mistakes, none of that. If they had the data, they credited it, and I haven't heard otherwise, so stop being trolls.

I also am 0n team ATI and prior to the crash frobozz had 3 to 5 clients running. now he may have added a masive farm since then I don't know. But I do know that there has been a problem with the stats since the crash. I have provided what I had for missing WU's to bruce and it appears even if you can show that WU's are missing it will be Ignored. And bruce really does not have verry good access to information. The only thing he can get is the last WU's submited by any given client which I did show 2 were missing but he has not got back to me since. Any way since you believe I am a troll and an iddiot it doesn't really matter now does it.

[J_M]

so stop being trolls.

Is this Pande Groups official stance then? If we notice anomalies in the recredit and have not been recredited ourselves we are trolls? Is this how an organization should treat it's volunteers?

If you represent the Pande Group it looks like it's time to find another project.

Actually his avatar is what a troll would looks like to me. lol

I'm wondering the same?

Ya, what I see is someone without a mod or PG tag entering what amounts to, for the most part, an intelligent discussion where we are attempting to get answers to begin calling names - at best with his 5,000+ posts and his name calling he is a wanna-be minimod and at worst HE is what he is calling us - a troll.

I find FAHReQ the most telling of all. This person was not even folding during the outage yet received a 'recredit' of 50,000 points on his 6th day and has averaged around 5k since. How can you have lost 50,000 points before you have even started folding? Surely PG must admit this recredit has been done incorrectly at this point.

I would like to see them explan that one as well?

Edit: BTW I fixed your link, it wasn't working correct.

Thanks - I should have checked it after I posted.

[djchandler]

Maybe these are the WUs which were re-credited by Stanford for situations where they had a team number but the user-name was undecipherable.

Bruce, you mean to say this is an acceptable practice, even where you have someone saying he/she didn't get credits, but it seems to be demonstrable the donor credited wasn't even active at the time of the stat server failure? Or is this speculation on your part alone?

This is exactly the kind of stuff I'm wanting to discuss. And this is exactly what is making some unhappy. I'm unhappy because our team all but lost its top producer for some reason, which coincidentally occurred on the same day as the stats server failure, just days after TeamUbuntu held what we called a FoldingJam. This occurring right after mat have really undermined our efforts. What you call coincidence I see as convergence. The two, are at the very least related temporally.

What really happened on 11/11-11/12? That's all we want to know. PG remaining silent is not helping. We deserve to know if there's a security issue.

[VijayPande]

I just wanted to give some info here. The recredit is finished. As noted in my blog, this one was much tricker than ever before, since (for the first time in almost 10 years) the flaw was in the stats server itself. During the time that the stats server was down, we did receive the completed WUs (the scientific data on the work servers) but we lost the data that associates that PRCG with the donor who uploaded it. For this reason, we had to statistically estimate, based on previous WUs, what the lost WUs were. Our thinking was that an approximation, based on the data we had, was better than no credit at all. However, this clearly has limitations and flaws and couldn't go as smoothly as previous stats recredits had in the past.

The issues with that server prompted me to upgrade the hardware (it's fairly old now -- over 5 years old) and the switch over to new hardware will help prevent this particular issue from happening again. I'm sorry for the problems here; I understand how critical points are to how FAH works and it pains me to see this sort of problem associated with FAH. However, we did the best we could with the data we had and have a plan to try to avoid this particular flaw from happening in the future, although it will take some time for new hardware to come and for us to transition over. My hope is that this will be completed some time in January 2010.

I'm very sorry for the hassles and problems here. Unfortunately, there isn't any more we can do about this incident, other than make plans to work to avoid this particular problem from happening again.

[kiore]

Thanks for that explanation Professor.
I think I got pretty close to what was lost (was running at 50k+ ppd at the time), it did look strange though, so now I understand. Points are fun and motivating, but personally I would have just worn the lost, with an explanation. So some winners and some losers probably, it should all average out over the long term.
Foldon everyone.
kiore.

[djchandler]

I just wanted to give some info here. The recredit is finished. As noted in my blog, this one was much tricker than ever before, since (for the first time in almost 10 years) the flaw was in the stats server itself...an approximation, based on the data we had, was better than no credit at all.

Thank you, Professor. That's an understandable explanation. Why it was so hard for someone to say there was a simple hardware failure leading to records getting disconnected is baffling to me. Here's hoping nothing happens between now and when the new stats server arrives.

As for the approximation of granting WU credits, that depends on which side one is observing the issue. Having checked your blog earlier today and finding nothing since the 23rd, it seemed the correct thing to do was come back here and ask questions once Bruce and others informed us the re-credit was complete. As for what fallout there may be, that remains to be seen. Hopefully those who feel wronged will be gracious enough to accept this mishap and continue supporting F@H.

As for myself, nothing will keep me from going on supporting these projects. Of all your F@H donors, I may have more to gain personally from your projects successes than anyone.

May you always be blessed with good health!

[Grandpa_01]

Thank you for responding Dr. Pande

I guess I am one of the unlucky ones. But I am happy to hear you believe all WU's were received. I personally would have like to have know about the issue much earlier rather than felling like my donation was unimportant and the PG could care less. You are dealing with John Q public here who would like to think he is doing some good. When a situation like this arises and a person notices something is wrong brings it up then gets horse whipped for it while the people who actually know what has happened stand by and do nothing for days, well I don't know that that is a good thing. Perhaps the PG should re think there policy of silence.

[k1wi]

I think the main thing here is the science got done. I don't know about anyone else, but I really couldn't care less if I lost a day's points (to this day I don't know if I lost or gained and I don't care!).

After all, I have enough problems keeping my systems up and running; how many work units have we each tossed mucking around with clients/OCing/set-ups? We all know how that pushes back the project... Consider us even.

I think it's probably time for the mods to lock this topic; it's been answered by Pande Group and any further issues are a new issue and should be in a new topic.

./K1wi

Edit: Speaking from inside Academia, I don't mind that Pande Group prioritise the science over customer/donor service.

[TheWolf]

Not sure were you been reading but it wasn't here the last 11 pages.
Looks like more bs to cover up the big screw up they made the way I see it.

For this reason, we had to statistically estimate, based on previous WUs

If they gave the amount or close to what you were making someone made a pretty good mistake in a lot of places.
Just say yall screwed up & leave it at that.

[bruce]

OK, time to close the discussion.