Folding Forum

Right so the issue is still there. Its just a bit more intermittent. It is likely that I will try and move some of our infrastructure to more robust machines tomorrow. Sorry for all the issues everyone.

Hardware fails sometimes, nothing last forever get er done...

msultan wrote:Right so the issue is still there. Its just a bit more intermittent. It is likely that I will try and move some of our infrastructure to more robust machines tomorrow. Sorry for all the issues everyone.

Hopefully you can come up with a failover system somewhere down the line, so that a failure in one system would simply activate the secondary to take its place while the first is fixed.

Thanks for keeping us updated.

So I am planning on starting the migration tomorrow morning at 10am. This means that the last stats update is gonna be at 9am PST. Hopefully, we should have everything moved and sorted out pretty soon.

It looks like the system decided to take a break at 1am Eastern time, but given the patterns of the last few days, I can't say I'm surprised. Thanks for keeping us updated and here's hoping the transition goes as smoothly as possible!

d

Stats should hopefully be fully back up. Please to the following URL for more details viewtopic.php?f=24&t=30285

I can verify that the 7 and 8pm Eastern updates came in and the values are incrementing as expected. Thank you again for the hard work!!

This is a little nit-picky, but I'm noticing a drastic offset on the team rankings, and there are a lot of users across most teams with "N/A" rankings. I've not seen this behavior before. Is this expected?

This is a little nit-picky, but I'm noticing a drastic offset on the team rankings, and there are a lot of users across most teams with "N/A" rankings. I've not seen this behavior before. Is this expected?

Could you please give me an example on the stats page? I can take a look in the morning by comparing the old and new dbs. I think the top few teams have the same ranking. The only users that dont have rankings are those with <100k aggregate points. Thanks a lot !

So, looking at my team, just as an example, I notice a couple things.
It is true that users with < 100000 points are not ranked, but I have never seen them show up on the report before.
Second, I see users who are apparently NOT ranked, but have > 100000 points, so I feel like this is not only offsetting the ranks as a whole, but also leading to an issue where people who should be ranked are actually not.

http://folding.stanford.edu/stats/team/69411

Examples of non-ranked users with > 100,000 points:
Remon5850
Lancelot
Andrew67
Dunimous
El_Magio
zinzuke
CMUndertaker
Sintares
Asztabart
CPLMayo
Neo_Cyrus
RotVik.MO
Jorn

As I remember, their was always some sort of cutoff in the past for the flatfile. I know a lot of 3rd party sites only give say the top 10k folders and top 3k teams.

But after looking at a couple of those folders, the one thing that strikes me is they are all inactive and havent by looks returned any workunits for a long while. Saying that I only looked at the first three on the list. But if the same holds true for the others, then their may be a cutoff for inactive folders.

But no. I looked at my own team, and all of my folders that did less than 100k in points are also NA. So to get ranked you need to donate enough to get 100k. Its not an unfair amount nowadays. An average gpu user can probably do that in a day. A cpu folder, a little longer. It's all about the commitment.

It's perfectly possible that the ranks are completely valid. A 10,000+ drop in rank over the course of three and a half days was just a bit startling
On a slightly different note, I wonder if the migration has broken the local web control's ability to pull stats in and display them:
https://drive.google.com/file/d/0BxGPu2 ... sp=sharing

Glancing at the flat files I store, keeping in mind that to save space I only track the top 50 teams on each pull, but DO get every team member from those 50 teams, not counting the team 0 listing), the flat files are indeed showing all folders with less than 100,000 points, so I would guess they made a choice to filter certain point levels out for the 'live' stats display. But that's only a guess based on the data I have access to.

d

Thanks for all the detailed reports everyone. I will get our front-end developer involved today to see if I missed some sub-routine's access to our system which is causing the weird ranking issue. I have already emailed them so it should hopefully get resolved fast.

Alright, all the ranking issues should be fixed now.