Page 1 of 1

software improvements , bad WUs

Posted: Sat May 02, 2020 9:33 am
by strombergFs
In case of a bad WU: is it possible that the next software automatically inform the scientist about a bad WU and sends the logs too - after that delete the WU and download a new one? The client waste time and some people dont recognize that they dont fold due to a bad WU. It would be nice if the software takes care about it automatically and ask for a new WU and delete the bad one..

Another Thing ( i do not know where to place it in the forum):
Supidly i changed one time my client name ...later i realized that the new name starts again with 0 WUs / 0 points. This is very sad, it is the same computer. Is there any possibility that someone can transfer the points/WUs from my old name to the new one ..?
In general, there should be a big warning in the software that a change of name result in loosing all statistics... I guess this warning would be a nice to have in the software..
The configuration page even say: the name can be changed any time....no warning that we loose all statistics.. this is a bit sad.
Thank you very much for any replay to this "problem".

Re: software improvements , bad WUs

Posted: Sat May 02, 2020 9:55 am
by PantherX
Regarding the bad WU, can you provide an example where it is "stuck" AFAIK, the domain decomposition issue happens on systems with large number of CPUs and generally speaking, Donors who own such kind of hardware, tend to keep an eye out and know their way around technology.

Please note that once a WU has been credited with a username/passkey/team combination, it will not be transferred at all. To clarify, you didn't "loose" any stats, they are still there with the original client name. Also, donor names are not unique. The only thing unique is the passkey which is recommended to use for security reasons and bonus points. Here's an article for passkey if you're interested: https://foldingathome.org/support/faq/points/passkey/

Re: software improvements , bad WUs

Posted: Sat May 02, 2020 9:56 am
by Neil-B
PantherX beat me to a response again :) - but since I've typed tit:

"Bad WUs" come in many different forms - most aren't actually bad (normally quite rare) and they are actually issues caused by issues at the client end (configurantion/hardware/anti-v/etc.) so just reporting and binning isn't for the most part appropriate as the issues at the client end need resolving ... Recently there has been a higher rate of issues (Bad WUs, Server, Installation, etc.) due to the rush to expand the infrastructure and workflows to respond to Covid-19 and I am sure lessons will be learned about this and some additional "folder assist" functionality may well develop in further updates and these will be prioritised by benefit to the science.

Transferring points does not happen ... Points are recorded when they are uploaded to the name/passkey/team at the time and this does not change ... There are numerous threads in the forums explaining the reasons why this is the case ... Statistics are not "lost" they simply remain where they were and you will need to manually add them together if you wish to keep track ... You can always look them up ... The one area this can have a real impact is if you change name with a passkey at which point the qualification period for the QRB is reset as it is for every username/passkey pairing.

I am sure the developer will take on board some of the issues seen over the last month and as/when possible include warnings and improve the onboarding process.

Re: software improvements , bad WUs

Posted: Sat May 02, 2020 10:15 am
by strombergFs
Yes, there was a gromacs fatal error...this WU could not be fold, no progress, but the slot was blocked. So the computer could not fold other WUs. All few seconds/minutes an error was written in the logs. luckily i saw that and asked in the forum for help. the scientist confirmed the bad WU, i deleted the WU after providing all logs and could go on folding new WUs.
See topic gromacs fatal error.

regarding the change of name: in the statistics my computer lost places because i changed the name. For FAH these 2 names are two different computers with two different statistics. And this must not be. A warning in the configuration tab that you start with a new name from zero would be nice. My computer would have now a better place in the statistics if i havent changed the name...which i regret.

Re: software improvements , bad WUs

Posted: Sat May 02, 2020 10:26 am
by PantherX
The GROMACS fatal error (viewtopic.php?f=16&t=34625) is more of an exception than normal. There are a total of 72 matches in this form which has records for several years (search.php?keywords=gromacs+fatal+error) so I would not personally stress over it. It is a good enhancement and can be on the to-do list but nothing close to urgent/critical.

Moreover, the current system can handle Bad WUs in most of the cases, refer to this topic: viewtopic.php?f=19&t=16526

The message could be an enhancement but that would be up to the development team to consider.

Re: software improvements , bad WUs

Posted: Sat May 02, 2020 10:30 am
by strombergFs
thank you both for your fast answers!
understood. Again thanks a lot.

Re: software improvements , bad WUs

Posted: Sat May 02, 2020 10:37 am
by PantherX
BTW, if you would like to, feel free to contribute to the Open Source for F@H. There are plans to make more of F@H open source in the future but this is a start: https://github.com/FoldingAtHome

Re: software improvements , bad WUs

Posted: Sat May 02, 2020 10:43 am
by Neil-B
In no way aimed as an attack on the OP ... indeed I thank strombergFs for catalysing the following "rant" which is truly aimed at the world in general - not the OP in any way ... "Thank You for giving me an opportunity to highlight something that has been bugging me for a while" :)

There are times I really wish that "Points" didn't actually exist ... For all the good they do (and arguably they do have a beneficial effect) they also cause so much/many issues/stress/arguments/anger that in many ways they are simply counterproductive, causing upset, concern, and a lack of focus on what FAH is really about - which is providing much needed and appreciated compute resource where it wouldn't otherwise be available. Vast amounts of effort are expended producing and worrying about something that is in the big scheme of things totally irrelevant :shock: :(

Points are awarded simply to acknowledge "Throughput" - to give people something they can see that recognises they made a contribution ... Human nature has this really perverse desire to turn everything into a game/battle and this is simply unnecessary ... Does it actually make any difference how many points one has? ... or whereabouts in the statistics "leader table" one is? ... Does that make any difference to the science or ones physical well being? - actually the reverse might be the case for both the previous.

Some will argue that Points competition is good that it drives innovation a gets/retains folders ... but it also makes many people unhappy and stressed - making them spend more on kit than they possibly should - getting people irate, angry, upset when instead they might actually be celebrating that they are contributing !!! - and at this time I think people could all do with a bit less stress and more happiness.

So "rant over" - I know many won't agree with what I have written, but I felt like saying it ... Stay Safe, Stay Well and Fold On :D

Re: software improvements , bad WUs

Posted: Sat May 02, 2020 5:32 pm
by Joe_H
One thing not mentioned yet here or in the original topic about the Gromacs error is that the reason the client did not fail out the WU and send a report. There is a known issue with the Linux version of the A7 folding core, for some errors it sets the wrong code. Normally after a WU failing like this a few times the client will dump the WU as faulty and send an error report. The same error if it had occurred on a Windows machine for example would not have tied up the CPU folding slot.

So the automatic handling of errors is already there. It is just that another component of the folding client processing does not generate the right error code on this specific OS to trigger that automatic process. Folding core development is separate, and not redone often. So eventually there might be a reason compelling enough to come out with a revision and this issue could get included in the new version.

Re: software improvements , bad WUs

Posted: Sun May 03, 2020 2:19 pm
by strombergFs
@Neil-B you are right, points dont matter at all. I agree! But at the end i guess most people incl. me want at least to know the progress they did. I have two computers running 24/7 with all CPUs just for FAH. I do nearly nothing else with them, so i give all CPUs to FAH. I can not compete with big hardware, it is just an old macbook and NUCi7.
However, i do my part. And i like to see every day if all is okay and if i finished some WUs. But with the change of names also my total of WUs from that "name changed computer" was reset. And i would like to be able to "delete" the old name and have my total of WUs / points..what ever ...displayed..it is the same computer. Now there is one name, which does not fold anymore...it is an inactive "computer" for the statistics and for my team.
Anyway it is not important, i just wanted to write a small wish. I do FAH to help, not to see points... So we can forget the wish :-) i continue folding...

@PantherX: that sounds interesting. I sadly do not have the skills to help. However, is there more to read about the project? The github page does not provide too much information or i overlooked them.. Does OpenSource also mean that there is maybe in the future the possibility that low power ARM based computers can also contribute? That would be nice. I guess there are many people with RaspPI , XUs and so on...they all could help too. I would like to fold with a solar powered XU4 ...it would make me feel even better...to power the NUC and macbook is for me not possible...

Re: software improvements , bad WUs

Posted: Mon May 04, 2020 4:26 am
by PantherX
strombergFs wrote:...@PantherX: that sounds interesting. I sadly do not have the skills to help. However, is there more to read about the project? The github page does not provide too much information or i overlooked them.. Does OpenSource also mean that there is maybe in the future the possibility that low power ARM based computers can also contribute? That would be nice. I guess there are many people with RaspPI , XUs and so on...they all could help too. I would like to fold with a solar powered XU4 ...it would make me feel even better...to power the NUC and macbook is for me not possible...
When it comes to reading, the main site (https://foldingathome.org/) and Wikipedia (https://en.wikipedia.org/wiki/Folding@home) are a good start. I am aware that when the website was migrated, it caused some content to be lost, some pages have no links within the site but can be searched. There are plans to work on that. When it comes to other devices, in the past, Android devices could fold but that was with Sony's partnership. When Sony stopped, the project came to an end but they do plan to revisit it in future. If the community contributes to the development of open-source code, it frees the primary developer to focus on closed-source stuff so let's see what happens :)