Folding Forum

A major overhaul of a service as complex as FAH SHOULD require it to be completely shut down until all of the compoonenets can be updated, tested, and prepared for the release of a new system. To avoid such a shutdown makes everything more complicated because (1) servers would need to quit assigning all outgoing WUs while those WUs which are already assigned are returned or expire. (and since most WUs have a 10 day, expiration date, that accounts for the first 10 days of the shutdown. (2) Somebody would need to visit all of the servers, install an updated version of Linux plus an updated version of the server code plus whatever in-process projects are ready to resume, and test that everything works together (which really can't be done without (3) a major group of volunteers who are prepared to beta-test the updated FAH infrastructure. Also (4) a new client would have to be distributed to all of the donors.

OK, now suppose FAH attempts to undertake the same changes without shutting down for a month. Doing the same changes one server at a time makes it a lot more complicated because now each work server, assignment server, stats server, client, (etc) has to be able to communicate with other servers/clients which might be running either the old or new version so everything not only needs to work when everything else has been updated, it still has to interface with components which have not been updated yet. Notice the variations of numbers in the "Version" column of https://apps.foldingathome.org/serverstats. Also notice the servers that are in Accept mode which still are collecting unexpired WUs where they still in the process of draining projects.

Inevitably, bugs will be discovered and fixed during this process while attempting the impossible task of keeping all the donors happy. Also, since from time to time some projects will be temporarily off-line, there will be periods when there aren't enough WUs available to be assigned to everybody.

FAH is in the midst of this major upgrade and considering everything that's going on, I think things are going as well as can be expected ... but still somewhat less than perfect. Moreover, development work is continuing on a new (as yet unannounced) client version which undoubtedly will be able to display the QRB correctly at all times and in all places.

I believe Development has been doing a yeoman's job of juggling many more tenpins/balls/torches/swords/etc. than can safely managed with a single pair of hands per juggler.

How easy would it be to just add the k-factor back into the psummary response json?

It doesn't need to come from the work servers - just put the hardcoded 0.75 in there? Wouldn't everything just work?

Surely easier than dealing with all the fallout on here.

Done.

rwh202 wrote:How easy would it be to just add the k-factor back into the psummary response json?

It doesn't need to come from the work servers - just put the hardcoded 0.75 in there? Wouldn't everything just work?

Surely easier than dealing with all the fallout on here.

jcoffland wrote:Done.

The displayed amounts for the Estimated PPD & Credit appear to be working for WUs that are coming from WSs running the older version 9.2.5 (e.g. 128.252.203.4) ... but not those coming from servers running 9.4.2 or above (e.g. 155.247.166.220).

I see. This will require another WS upgrade. The necessary change is in the next WS release but it will take a few days to roll this out.

Joseph, do you plan to update FAHControl/FAHClient ?

Glad I finally found this. I've spent the last hour trying to figure out what might be wrong with my PC. Timing was poor since I just applied the August Windows 10 patch.

Kebast wrote:Glad I finally found this. I've spent the last hour trying to figure out what might be wrong with my PC. Timing was poor since I just applied the August Windows 10 patch.

Same here, it coincided with me swapping PSUs between my two machines, and thought maybe I did not connect all the cables correctly

Is there any update as to when FAHControl will show the correct PPD?

HaloJones wrote:Is there any update as to when FAHControl will show the correct PPD?

Until the developers get around to it, nobody knows when it will be. It will happen one day, that's the best answer you are likely to get.

The developer is working on it. However the fix has to be rolled out to a number of different servers that are actively handling WU assignments and stats updates.

An attempted quick fix did not work out.

Ultimately a full fix may require an updated client, that will take longer.

It would appear things are improving. This morning FAHClient reported the QRB correctly for all the slots I currently have running for the first time in a couple of months. There appears to be only one active Work Server at Sloan-Kettering still on 9.2.5 but I understand that this is dynamic and may change with projects coming on-line.

Your'e using the wrong words. There are two assignment servers: 65.254.110.245 (at wustl) and 172.31.11.240. They ASSIGN you to an active work server. Then it's the work server that distributes WORK. You mean to say there is only one active Work Server at Sloan-Kettering on ...

bruce wrote:Your'e using the wrong words. There are two assignment servers: 65.254.110.245 (at wustl) and 172.31.11.240. They ASSIGN you to an active work server. Then it's the work server that distributes WORK. You mean to say there only one active Work Server at Sloan-Kettering on ...

Sorry, I should know better than to post anything before finishing my first Coffee ...

I'm seeing what appear to be the right numbers for PPD and Estimated Credit from WU's coming from 128.252.203.10 and 155.247.166.220. I don't have anything at the moment from 128.252.203.4.

It certainly appears that FAHControl has been fixed, at least based on how my system is performing the last 24 hours now