Changeover Inquiry

[MgKnight]

As we draw ever closer to what will likely be a change of protein on Tuesday I am relatively certain that I am not alone in wondering if this change will go smoother than the last few. It appears that Eleana, Howard and company have pretty much ironed out the bugs we have thus encountered and they are to be commended for that.

Will we be looking at a server shut down at 11:00 EST again? Will the down time be anything less than it has been in the past so that changes can be accomplished prior to the end of the work day? Will we again be working off of a single update server or have mirror sites been offered up and set up? If mirror sites have not as yet been offered up and set up is it too late to consider this?

It sure would be nice if I could get my sneakernet farm changed over on transition day instead of being a day late again.

[shortfinal]

One thing I would like to add is here in the US Monday is a holiday (Labor Day) and I run the client on the computers at work with '-i f' (nonet) because they're behind a firewall. So, on Tuesday morning there will be a lot of data to upload.

Normally, after a weekend of running it takes until 10:30 AM to upload all the data. With the 3 day weekend the upload will go on into the early afternoon.

A protein switch would cause me problems because I have to do the switch manually. Again, this is because of the firewall. Could the switch be done in the late afternoon or on Wednesday?

Shortfinal

[Brian the Fist]

We have Labour Day here as well. So it will be Wed., yes. We have no mirror sites still. Anyone with a T1 line or better and who's box is on 24/7 who wants to be a mirror, please feel free to contact us. All you have to do is set up the proxy daemon program on your machine (and tell it to download ALL OSes obviously) and give us the URL and we can make you an official mirror. I prefer a login account on the box too so I can check everything is working properly but this isnt compulsory. This would be greatly appreciated by us as well as your fellow folders.

Thanks.

[rsbriggs]

If anyone out there could set up a bitorrent download site, that would be a big plus, too. EVERYONE would be able to download the client within a few minutes, rather than having the typical two-day download jam-up and clients aborting because the server is swamped...

[Stardragon]

Bit torrent sites aside,as we are currently still without mirror sites, I would kindly request that you tried to upload all your buffered work before Wednesday's switchover.
We're going to need all our bandwidth for teh actual update, and in the long run, it's just to make YOUR life easier

[pointwood]

A possible mirror!

I asked in our forum and a guy name Pjort said he might be able to provide a mirror - you can find his email addy and ICQ # in this thread:
http://arstechnica.infopop.net/OpenT...1&m=2860985285

/me crosses fingers

[pointwood]

What kind of new protein can we expect? A new fast one or?

[Keller]

A similar one to our current, "small and well characterized" (reading the news would answer some questions )

[FoBoT]

Originally posted by Keller
A similar one to our current, "small and well characterized" (reading the news would answer some questions )

so interpret for us non-science types, howard posted
"be switching to Protein L"

how does that translate to
"small and well characterized"

[IronBits]

From the 'news section'

08/13/2003

- Our new protein is underway, we are now studying the folding of Protein G, soon to be followed by Protein L. These are small, well-characterized proteins that have been studied by other methods, and this will allow us to compare our 'folding movies' to other, more tried and tested studies of protein folding.

[FoBoT]

oh, i only read the newest entry

nevermind


move along, nothing to see here

[tpdooley]

.. While there's a few more days to go.. how did the top 3 or 5 do versus this same protein worked on the other "tried and tested" techniques?

During the phase I comparisons - we were consistantly in the middle of the pack. How much closer to the head of the pack have we got? And unlike some of the other techniques - we were consistent.. i.e. not at the head of the pack for one type of protein and the tail end of the pack for another type.

And if the results aren't available yet.. when can we expect to see the good news from Drs Hogue and Feldman?

[DGROMS.com]

Howard,
I will be able to provide a mirror with 2Mb download bandwidth for the changeover. I will get the daemon up and running tomorrow when I go to work and I will PM you with the details of the URL. At this stage I will create a URL df.dgroms.com, but I will tell you when it is up and running

Regards
DGROMS.com

[FoBoT]

thanks DGROMS.com

[IronBits]

DGROMS.com
Thanks!

[Brian the Fist]

Originally posted by tpdooley
.. While there's a few more days to go.. how did the top 3 or 5 do versus this same protein worked on the other "tried and tested" techniques?

During the phase I comparisons - we were consistantly in the middle of the pack. How much closer to the head of the pack have we got? And unlike some of the other techniques - we were consistent.. i.e. not at the head of the pack for one type of protein and the tail end of the pack for another type.

And if the results aren't available yet.. when can we expect to see the good news from Drs Hogue and Feldman?

We will not be analyzing results of the protein G and L test fully until we've completed sampling of both. However our goal here is different than phase I - we are not trying to guess their structures, but rather predict how they fold. The best structures found (2.5 A or so) are close enough to be within experimental error of the true backbone structure of the protein, so this means the corresponding folding movie COULD represent a possible path the protein folds along. What remains to analyze is whether the pathways we found agree with experimental evidence.

After this test, we will likely return to 'true' structure prediction tests, not using knowledge of the native fold of the protein to influence things. This will allow us to test out some new energy functions and algorithm ideas.

[tpdooley]

okay.. time to get this next update, steam through billions of predictions.. and then start creating results we can compare to other project's results..
I'll be interesting to see how we're doing for both sets of testing..

[PCZ]

Howard

If you require another mirror I can help out.
I have a shed load of bandwidth and spare servers.
As I have never provided a mirror before I will need
instructions.
Is it just a case of installing the autoupdate daemon ?

[Darkness Productions]

Howard -

Someone (possibly me) may have asked this once before, but is there any chance (after this protein update) to allow the update proxy to pull from other sources as well? It'd be nice to be able to offload some of the proxies to the new mirrors, and thus guarantee at least *some* decreased bandwidth on your end.

dp

[Paratima]

Originally posted by PCZ
Howard
If you require another mirror I can help out.
I have a shed load of bandwidth and spare servers.
As I have never provided a mirror before I will need
instructions.
Is it just a case of installing the autoupdate daemon ?

Let me know also. I may soon be the recipient of some bandwidth.

[PCZ]

Paratima
I worked it out myself.
1: Download the auto update daemon
2: Edit the config file, specify a download directory and select support for all clients.
3: Set up a web server to serve files from the download directory.
4: Create a DNS record for the web site.

Quite simple really.

[Paratima]

Tanks, PCZ! Won't know about the bandwidth for another few days, anyhow.
But it's good to know!

This means you'll mentor me if I screw it up, right?

[DGROMS.com]

Originally posted by PCZ
Paratima
I worked it out myself.
1: Download the auto update daemon
2: Edit the config file, specify a download directory and select support for all clients.
3: Set up a web server to serve files from the download directory.
4: Create a DNS record for the web site.

Quite simple really.

That it is. I set mine up again last night and allow it to download the 712Mb of update files. I should be ready to server later on today. The server is at
df.dgroms.com -> 61.95.87.13

[PCZ]

The only problem I had was when I uncommented ALL and started it up the program didn't download anything.
I had to uncomment all the numbers.

[DGROMS.com]

Originally posted by PCZ
The only problem I had was when I uncommented ALL and started it up the program didn't download anything.
I had to uncomment all the numbers.

That's weird, did you leave the * on that line and only delete the #, because I have never had that problem before and I have always used the daemon to update my clients

[PCZ]

My Fault typo

[Brian the Fist]

Again, if you want to be added to the pool from which all the clients (and other proxy daemons) grab their files, you must e-mail us at trades@mshri.on.ca once you've got the daemon up, and give us the exact URL where the files are located so we can 'hook you up'. Please also give us your approx. bandwidth and availability (i.e. 24/7?) We more or less depend on the mirrors being up whenever we update - if one is down, it could result in delays as clients try to reach it and must wait for timeouts, so please only volunteer if you intend to keep your machines available and check that the daemon is still going once in a while and hasn't mysteriously died If you create us an account on your box we can do this last part ourselves if you wish.

[pointwood]

Howard, have you contacted Pjort from us (Ars)?

[DGROMS.com]

Howard,
My Server is up and running at df.dgroms.com. I have sent you an email too.

Regards
DGROMS.com

[iggy]

I was just wondering if all those distribfold-patch-xxx-win9x.exe are necessary.

It seems that deamon (for Windows) is downloading everything from xxx=106 onwards.

I was checking on what files are really needed with two previous version of the client: only filelength.txt and 114 (plus 115) get pulled from the server. So, my guess is that with the new update, a new 116 patch will be issued, the deamon will get it, and the current client will update itself with that one.

I understand that there might be some folders who have been absent from the scene for some time, and that having all updates available can allow them to just restart that old client that will automatically update itself, but on the other hand, all those old updates do consume a lot of space (about 50 MB per OS).

Is it possible to delete some of those old patches if the update is needed only for the current version of client, and which one should be left in that case?

I'm trying to set up simple server for our team, but space is at premium (50 MB from my ISP, but huge bandwidth)...

Any info on this would be much appreciated.

[Stardragon]

Since updates are strictly incremental, and it is possible that some people are missing one or more required updates, all the updates are in fact essential. If any old updates are missing, a user will not be able to update for example from patch 112 to patch 116. Any differences greater than just on update need to be installed, especially since many updates are based on changes implemented in preceding updates.

So if you exclude any updates, you are doing so at the risk of having users unable to properly update - personally I would advise against that, unless you are 100% certain all users are currently fully up to date.

[iggy]

Thank you very much for your reply, Elena.

As all our members are fully up to date, and 99% are running WinXX, I'll provide as many newer updates as space allows.

[Brian the Fist]

We could add a flag to it for the next version for 'start at release # xxx', this would solve your problem, at least until a few more releases...

[iggy]

Howard,

that would be fantastic!

In that way I could easy provide updates for some other OSs for my team, as well.

I guess every little helps in keeping DF servers as free as possible!