DF servers

[grunter]

The DF servers are back up but are getting a pounding from buffered structures. I've got most of my machines working offline until things smooth out. I knew this would happen this is one of the fastest protieins I've seen.

 

[gistech1978]

you know i thought that when the servers are down the client would keep folding.
came into work this AM to find my 1.6 and 1.8 idle.
grrr.

 

[grunter]

I don't know what causes that. Some of mine keep going and some shut down.

 

[cbcrus]

Yes, most of my folders died here as well. Swtiched them to folding offline for time being and now am slowly allowing them back online one by one. The servers seem to be doing ok now as they have been accepting my uploads with no time out errors.

Always fun when you get to crunch a fast protein.

 

[grunter]

Whew that dump felt good.

 

[gistech1978]

again this AM
i figured everything was kosher yesterday afternoon evening
turned off no-net
wake up to find the laptop idle and 1 comp idle @ work.
man this is starting to irk me a tad.

 

[kmmatney]

My computers haven't gone idle, but they do pause for periods of time when trying to upload to the DF servers. Only one of my machines at work was able to download the protein update, and I couldn't even download the update manually from the servers. I ended up copying the entire folder from my only updated machine to all the others, in order to get them working on the new proteins. They all pciked up where the first machine left off - with 1000 structures already processed. I hope the client doesn;t try to submit all of my duplicate work units...

 

[xpblackdragon]

I don't have problems with this new protein the problem I have is to upload
the data the servers can't handle it:disgust:

 

[gistech1978]

Originally posted by: xpblackdragon
I don't have problems with this new protein the problem I have is to upload
the data the servers can't handle it:disgust:

no they cant.
i have well over 2000 gens buffered b/w home and work.
its taking forever to upload them. it uploads about 30 at a time then stops.
:|

 

[myocardia]

You know, what really pisses me off is that our computers are doing the work, saving them untold millions. Why can't they invest a little more money in servers (or bandwidth, if that's their problem)? It's not as if they need supercomputers just to collect data!:| I've already sent them an e-mail, telling them that if they want me to keep folding their proteins, they had better do something, and soon, or I will be finding a different DC client. You know, if enough people would do the same, maybe they'll get their a$$ess into gear. I can't believe that it's still not solved. I'm getting pissed!

 

[gistech1978]

Originally posted by: myocardia
You know, what really pisses me off is that our computers are doing the work, saving them untold millions. Why can't they invest a little more money in servers (or bandwidth, if that's their problem)? It's not as if they need supercomputers just to collect data!:| I've already sent them an e-mail, telling them that if they want me to keep folding their proteins, they had better do something, and soon, or I will be finding a different DC client. You know, if enough people would do the same, maybe they'll get their a$$ess into gear. I can't believe that it's still not solved. I'm getting pissed!

yeah, its starting to become more of a hassle than its worth.
im contemplating running another client for a while too, whats the point of buffering gens when it would take well over 2 days to upload all i have right now.

 

[xpblackdragon]

I have 3678gens to upload on dsl and is taking for ever

 

[kmmatney]

Somehow, my work has been uploading OK (for the most part), but it has been a general pain updating all of my computers, since auto-update didn't work too well.

It could be my imagination, but disconnecting the computer from the internet seems to make DF run faster...

 

[gistech1978]

Originally posted by: kmmatney
Somehow, my work has been uploading OK (for the most part), but it has been a general pain updating all of my computers, since auto-update didn't work too well.

It could be my imagination, but disconnecting the computer from the internet seems to make DF run faster...

i think youre right.
as slow as these uploads have been. lets say an average of 15 seconds for each gen.
thats 15 idle seconds * 250.
but it begs the question, does that equal the amount of time it takes to upload en masse?

im getting rather peeved at the lack of communication or updates or anything to try and remedy this problem from the staff working DF.
they are more worried about FTP access than fixing the problems we are having apparently. yeah. thats all we need are even more people running the client right now. it would be great, if they could keep up, but they cant.

 

[mikela78]

15 seconds? I just timed two uploads, one generation each. One was 30 seconds, the other 3 minutes. That may be longer than it takes to fold a generation.

Today a new client became available that makes 200 structures per generation. I'll try it out as soon as my client gets to 250 generations. If people start adopting it, the server problems will be greatly reduced.

 

[myocardia]

Originally posted by: mikela78
15 seconds? I just timed two uploads, one generation each. One was 30 seconds, the other 3 minutes. That may be longer than it takes to fold a generation.

Today a new client became available that makes 200 structures per generation. I'll try it out as soon as my client gets to 250 generations. If people start adopting it, the server problems will be greatly reduced.

Well, my current rig folds an entire generation in a minute or less most times, but what's the point of even running their damn client for them, when it takes an average of 1-2 minutes just to upload. And, have any of you considered the fact that this new protein's data is half of what the last generation's data was? I think today was my last day having anything to do with DF. Sorry guys, but I was doing this for the enjoyment of it, and it hasn't been enjoyable for quite some time now, because of whatever sub-moron is calling the shots at DF. I'm going back to the DC project that I was going to go with 6 months ago. See you guys later.

 

[myocardia]

Originally posted by: mikela78
15 seconds? I just timed two uploads, one generation each. One was 30 seconds, the other 3 minutes. That may be longer than it takes to fold a generation.

Today a new client became available that makes 200 structures per generation. I'll try it out as soon as my client gets to 250 generations. If people start adopting it, the server problems will be greatly reduced.

Oh, I forgot to address the bold portion. The client you're referring to won't change anything at all, for this reason: it will only take twice as long to do 200 structures, but doubling the structures folded also doubles the amount of data that has to be uploaded back to the servers. So, it's going to be identical, whether they use a 100 structure generation or a 200 structure generation.

 

[mikela78]

I'm not convinced doubling the number of structures per generation also doubles the amount of data that has to be uploaded. Each generation minimizes the energy of the best structure found, and this change simply doubles the number of candidates. You're still uploading one protien per generation. Also, this client update was in direct response to their overloaded servers, so we know the development team thinks it has an effect.

I'm using the new client on one of my boxes, and it works fine but I'm still looking for confirmation that I'm awarded twice as many points per generation.

 

[myocardia]

Well, it doesn't concern me anymore, so I'm not gonna worry about it. It's already been deleted from my hard-drive. I gave them time, and ample warning, and they blew it, in my opinion.

 

[cbcrus]

myocardia - I hear you, the server problem has been frustrating. Sorry to see you depart.

It appears that the newest client (200 structures/generation) must be reducing the load on the servers. My backlog of 2500+ generations is down to less than 300 this morning. I was beginning to wonder if they would ever get uploaded.

 

[kmmatney]

The problem is that the cleints aren't auto-updating to start calculating 200 structures/generation. I'm manually updating mine, for the benefit of all, but it is a pain. :|

Edit: I wonder how much work grunter has backed up.

 

[cbcrus]

Yipes! Just looked in my rear view mirror and saw that kmmatney is gaining at an incredible rate!
Time to roll up the windows so I don't eat too much of his dust. :beer:

 

[mikela78]

For the curious, the new 200 stuctures/generation client does not award twice as many point per generation, at least not with the client I downloaded.

This is a little disconcerting. Conscientious users have helped relieve the worst of the server traffic by using a new client which actually penalizes their benevolence. And what of science? If they can't tell the difference between the 100-structure and 200-structure clients for the stats, they've probably lost some of their control variables and can no longer judge the effectiveness of their algorithm.

 

[kmmatney]

Yeah, I'm making an all out charge with this protein. All but 3 of my systems are running DF, and I'm running 24/7 on machines that normally the DF client through my duplex screensaver.

I had a 500 generation dump yesterday (they all went through) so my numbers are a little inflated right now. I'm gonna try to keep my full herd at DF at least until the end of this protein.