Distributed Folding Project server trouble explained

[Brian the Fist]

Hello all,

I just wanted to take a moment to point out the cause of the server trouble. I am shocked that everyone is so upset over this but anyways because the new protein builds about 5x faster than the first one, our puny server cannot deal with the load and cannot handle all the uploads it is receiving, hence the Unable to Contact Server messages some of you have been receiving. It has around 25 uploads coming in simultaneously non-stop (as opposed to 3-4 beforehand).

Unfortunately we do not have the funding right now to get a newer more powerful server (I've found a nice Dual-P4 Xeon server available for about $2500 US) so what I intend to do is change the upload period to 25000 instead of 5000 for this protein. If those of you running on many CPUs apply this patch (when I provide it) it will hopefully ease the load on the server. If not, I can assure you that when we switch to the next protein I'll choose a larger one and it should work as smooth as it was before the new protein.

Incidentally, I've fixed two or three bugs with the auto-update - this was the first large scale test of it. Of course the auto-update that you got (one way or another) won't have these fixes so the bug is still there, but in the NEXT next auto-update it should work quite well I hope. In other words You will probably have to do one more manual update to get the auto-update fixes.

Anyhow, I will attempt to get a version out Monday with these auto-update fixes and with the 25000 upload period and hopefully that will improve things.

Sorry for the inconvenience and we hope you will be patient while we do our best to improve things.

Please pass this info on to other Forums if you participate in some others.

[IronBits]

Welcome to our humble home! Thanks for taking the time to tell us what's going on! That's half the battle right there.
I look forward to your bug stomping efforts so we can get back to work ;)
We HATE when our computers are not producing work, sitting idle not doing anything, when they could be working to solve problems, kill bugs, wipe out diseases and such.
When a client or project has a snafu, many of us have a way to switch projects instantly, until the problem is resolved.
If you provide a solid client, like you did have in the 1st round, I assure you that your project will never suffer from lack of computing power.
Oh, and we are all quite proud of the beasts we build to do this with! :D

[guru]

I have a Sun Ultra 2 dual 300 2MB cache with 1 gig of ram sitting around collecting dust. I may be able to loan it out to help with the project. I also need a solaris 2.6 client. If you need a system to compile it on let me know. I can setup one up for developement work.

guru

[Paratima]

YEAH! We LIKE this project!

Uh, :confused: I seem to find myself a bit short on Dual P4 Xenon thingies. :p

However, if there was an address to mail a check to, or a Paypal account or sumpin, I could probably kick in a couple bucks for the cause. And I bet I ain't alone.

[Dyyryath]

Heh, I had a sneaking suspicion that the shorter WUs were the problem. Going with a longer set of structures will temporarily alleviate the problem, but it's not going to help much as more and more people join the project.

How long will it be before they have 10,000 users? Or 20,000 users? Hardware upgrades are obviously in order, but I certainly understand budget restrictions. I have to deal with them everyday at work.

So, how can we help out here? I really like this project and I'd like to see this fixed. The guys over there are obviously responsive. Paratima has one idea. I'd be willing to donate something to the cause as well. Guru has another. Some of use probably have hardware that could be donated. I'd also be interested in getting a little more information about their current setup over there.

As many of you know, I'm the Online Systems Administrator for a newspaper and my job is fixing these types of problems day in and day out.

Sometimes things can be streamlined to allow existing systems to do much more than they currently are. We've got a dual P3-600 in service as a webserver (serving dynamically generated pages from a database no less) that sees an insane number of hits per day due primarily to very carefully written software and a highly tweaked configuration.

Distributing the load across several inexpensive (possibly donated) servers could be an option as well.

Without knowing how they are currently handling the load I can't really say whether any of these things are possible. They may be doing the very best that they can with what they have already. However, if they'd be interested in sharing the details with us (or even a small group of carefully selected volunteers with some expertise in certain areas) we may very well be able to help them work around their funding problems...