Auteur Sujet: Discussions on Formula Boinc [in English]  (Lu 490 fois)

0 Membres et 1 Invité sur ce sujet

Icecold-Team Anandtech

  • Messages: 6
  • P'tit Nouveau
  • *
Discussions on Formula Boinc [in English]
« le: 07 octobre 2020 à 02:05 »
English:

I do not speak French very well, so I will include a Google translate as well and hopefully it is clear enough.  I was wondering if you have been following what has been happening with the ODLK project and what your opinion is of it.  Basically, the project has been run out of tasks for several days while large bunkers were built in anticipation of it being chosen as the Formula BOINC Sprint and the project scientist is basically in a standstill with running the project.   Currently the site is up and down and there haven't been tasks available to crunch for days.  The project scientist was on their forum wondering what was going on.  Regardless of what is thought of the bunkering tactic, etc. it seems like it really highlights the need to be more mindful of what projects are eligible to be the sprint project and maybe taking away voting.  It would not take much to basically shut down a project like ODLK or T Brada by grabbing all available tasks, so I'm hoping maybe something could be done to improve that aspect in the Sprint.  This isn't my goal to complain about bunkering or other teams, but just to bring attention to the harm caused to small projects by how the Sprint currently functions.  I really appreciate any input you may have.

Français:

Je ne parle pas très bien le français, donc j'inclurai également un traducteur Google et j'espère que c'est assez clair. Je me demandais si vous aviez suivi ce qui se passait avec le projet ODLK et quelle en était votre opinion. Fondamentalement, le projet a été à court de tâches pendant plusieurs jours tandis que de grands bunkers ont été construits en prévision de son choix comme Formula BOINC Sprint et le scientifique du projet est fondamentalement dans une impasse dans l'exécution du projet. Actuellement, le site est en marche et il n'y a pas eu de tâches disponibles depuis des jours. Le scientifique du projet était sur leur forum et se demandait ce qui se passait. Indépendamment de ce que l'on pense de la tactique de soutage, etc., il semble que cela souligne vraiment la nécessité d'être plus attentif aux projets éligibles pour être le projet de sprint et peut-être retirer le vote. Il ne faudrait pas grand-chose pour fermer un projet comme ODLK ou T Brada en saisissant toutes les tâches disponibles, alors j'espère que quelque chose pourrait être fait pour améliorer cet aspect dans le Sprint. Ce n'est pas mon objectif de me plaindre du bunkering ou d'autres équipes, mais juste d'attirer l'attention sur les dommages causés aux petits projets par le fonctionnement actuel du Sprint. J'apprécie vraiment toute contribution que vous pourriez avoir.
« Modifié: 08 octobre 2020 à 14:46 par ousermaatre »

cougarpelardou

  • Messages: 1874
  • Boinc'eur devant l'éternel
  • *****
  •   
Re : Re : Formula BOINC 2020
« Réponse #1 le: 07 octobre 2020 à 08:25 »
You're welcome IceCold 

The Alliance Francophone is more of a long distance running team than a sprint team.  :frenchy:
Check about the bunkering with team like The Scottish Boinc Team, they are very clever with bunkering for the FB'S sprints

And Beware #covid19



« Modifié: 07 octobre 2020 à 08:28 par cougarpelardou »

Icecold-Team Anandtech

  • Messages: 6
  • P'tit Nouveau
  • *
Re : Re : Formula BOINC 2020
« Réponse #2 le: 07 octobre 2020 à 13:33 »
You're welcome IceCold 

The Alliance Francophone is more of a long distance running team than a sprint team.  :frenchy:
Check about the bunkering with team like The Scottish Boinc Team, they are very clever with bunkering for the FB'S sprints

And Beware #covid19



I appreciate the response!  I started with the other teams, and the response was not well received.  Albeit, I'm sure I probably could have been more tactful in how I presented it as well.  :D I was more curious if this team had any ideas on how Formula BOINC could be improved in regards to the Sprints to prevent some of these issues.  With a project as small as ODLK all it takes is one person or a few people to create a bunch of BOINC instances a week before the Sprint is announced and they can take the project completely 100% down without any tasks processing. It's not just ODLK, there are several small projects that are not capable of hosting a Sprint and it causes great detriment to these projects to be included in them with all of the speculative bunkering, etc. that it causes to happen to them.

cougarpelardou

  • Messages: 1874
  • Boinc'eur devant l'éternel
  • *****
  •   
Re : Re : Formula BOINC 2020
« Réponse #3 le: 07 octobre 2020 à 14:28 »
100% Ok with you. I check this two months ago with 3 teams ...

And I intend to do the same  But it didn't seem fair to me. So I cancel :D

modesti

  • CàA
  • Messages: 17870
  • Boinc'eur devant l'éternel
  • *****
  •   
    • Brocantes en Bourbonnais
Re : Re : Formula BOINC 2020
« Réponse #4 le: 07 octobre 2020 à 14:58 »
Welcome Icecold :hello:

It's true that a couple of projects are too small to support a Sprint. Maybe the Formula Boinc should be re-thought, again... Suggestions are welcome :)

Traduction:
C'est vrai que certains projets sont trop petits pour supporter un sprint. Peut-être que le FB devrait être repensé, encore une fois... Les suggestions sont les bienvenues :)

Viendez chez nous, cause qu'on est les meilleur(e)s :D


In memoriam Jip

cougarpelardou

  • Messages: 1874
  • Boinc'eur devant l'éternel
  • *****
  •   
Re : Re : Formula BOINC 2020
« Réponse #5 le: 07 octobre 2020 à 15:44 »
Certains projets comme Gerasim limitent à 1024 par utilisateur le nombre de tâches en cours. Cela limite la rétention industrielle.
En fin d'année, quand le nombre de projets possibles de réduit, d s équipes comme les Écossais commencent à faire de la rétention  plus d'une semaine avant sur 3 ou 4 projets.


Some projects like Gerasim limit the number of current tasks to 1024 per user.  This limits juge bunkering.
 At the end of the year, when the number of possible projects decreases, teams like the Scots start to bunker more than a week before on 3 or 4 projects.
« Modifié: 07 octobre 2020 à 15:49 par cougarpelardou »

Icecold-Team Anandtech

  • Messages: 6
  • P'tit Nouveau
  • *
Re : Re : Formula BOINC 2020
« Réponse #6 le: 07 octobre 2020 à 17:50 »
Welcome Icecold :hello:

It's true that a couple of projects are too small to support a Sprint. Maybe the Formula Boinc should be re-thought, again... Suggestions are welcome :)

I appreciate the response!  My suggestions would be to remove the handful of projects from the sprints that cannot handle them, and also to quit allowing voting.  In this sprint there were a ton of votes for ODLK which can't handle a sprint as well as SETI and Enigma that are not active projects, so it's clear many teams are voting as a joke or not in good faith for what projects can handle it. I was happy to see that Numberfields was chosen instead(even though it had less votes) since it seems much more likely to properly handle a sprint.

[AF>Libristes] Dudumomo

  • Messages: 6052
  • Boinc'eur devant l'éternel
  • *****
  •   
Re : Re : Formula BOINC 2020
« Réponse #7 le: 08 octobre 2020 à 03:35 »
Modesti, correct me if I'm wrong, but we are supposed to contact the admin and ask them if they agree with a sprint, no?
This should provide visibility to the project admin and help him to reduce the stress on his server.

Icecold, at the same time, the project admin can put a limit of the numbers of units a user can stock. This will avoid the issue of bunkers. (But not the surge of users)

In one hand, organizing sprint is not always easy and high participation is very positive for the community (Teams involved etc..) and normally the project as well, but in the other hand I know we had some "Crash tour" in some cases, where the server could not handle the load...we usually take it positively as it still provide a boost in the progression of the project, yet it can also impact the project more seriously.

One way could be indeed to change some of the rules of the Formula BOINC and others competitions such as:
- Always ask admin approval
- Remove the projects that are known weaker (based on their hardware I suppose, or simply through the admin approval)
On the vote, it's not really an issue, if the list of projects to vote can all handle the load.

Another way, that I prefer, is to further support the project admin by implementing some temporary load balancer (Additional servers to handle the load, download/upload or else). But this become a technical discussion with the BOINC Server platform. I've raised this suggestion in May this year on BOINC forum, but probably few people are interested only.
https://boinc.berkeley.edu/forum_thread.php?id=13692#98398

Keep the crunch up ;)  :hyperbon: :hyperbon:
News & Tutorial on how to host your server: http://freedif.org

Icecold-Team Anandtech

  • Messages: 6
  • P'tit Nouveau
  • *
Re : Re : Formula BOINC 2020
« Réponse #8 le: 08 octobre 2020 à 04:34 »
Modesti, correct me if I'm wrong, but we are supposed to contact the admin and ask them if they agree with a sprint, no?
This should provide visibility to the project admin and help him to reduce the stress on his server.

Icecold, at the same time, the project admin can put a limit of the numbers of units a user can stock. This will avoid the issue of bunkers. (But not the surge of users)

In one hand, organizing sprint is not always easy and high participation is very positive for the community (Teams involved etc..) and normally the project as well, but in the other hand I know we had some "Crash tour" in some cases, where the server could not handle the load...we usually take it positively as it still provide a boost in the progression of the project, yet it can also impact the project more seriously.

One way could be indeed to change some of the rules of the Formula BOINC and others competitions such as:
- Always ask admin approval
- Remove the projects that are known weaker (based on their hardware I suppose, or simply through the admin approval)
On the vote, it's not really an issue, if the list of projects to vote can all handle the load.

Another way, that I prefer, is to further support the project admin by implementing some temporary load balancer (Additional servers to handle the load, download/upload or else). But this become a technical discussion with the BOINC Server platform. I've raised this suggestion in May this year on BOINC forum, but probably few people are interested only.
https://boinc.berkeley.edu/forum_thread.php?id=13692#98398

Keep the crunch up ;)  :hyperbon: :hyperbon:

  :D :hyperbon: :hyperbon: :hyperbon:  :D

I really appreciate the reply! Everybody has been super welcoming towards me here, which I really genuinely appreciate!

As far as I know the project admin is not normally contacted.  That has been a point of contention on a few sprints recently.  That could be for many reasons - maybe the project admin was sent an email and didn't see it?  I'm really not sure, but there have been several sprints where the project admin has no idea it was happening. Making sure they are contacted and are able to handle the sprint would be super helpful.

As far as the vote, I guess ultimately that would be something that would be decided on what the goal is on Formula BOINC.  If it's 100% just based on getting exposure to smaller projects and encouraging extra crunching - you are likely correct and eliminating the small projects that are unable to handle the Sprint would be most likely sufficient. If there's any desire to keep this as a fair competition as well as encouraging extra resources put towards those small projects it is vital that the vote is ended. Keeping it a fair competition would encourage extra participation as well.  I would suggest you look at the vote history and see how often there are teams voting for the same project, how much they have pre-bunkered, etc.  Having it be voted on rather than selected independently encourages extra "scheming" to try to win the vote across teams, and doing detrimental practices by bunkering a ton of tasks that may or may not be completed, etc. which can be detrimental to projects.  ODLK was an extreme example - the project was run completely out of tasks for several days based on some teams knowing it would receive the most votes, but the precedent has been(somewhat) set that it's a viable strategy to grab as many tasks, etc. ahead of time if you know it may be selected.  Eliminating the vote would eliminate that aspect, which so far seems to be a very negative aspect.

If there is any way to assist projects have a temporary load balancer that would be great, and I would be happy to donate resources however necessary whether it's PC resources, or a donation towards cloud resources, etc.

I really appreciate your response!

 :D :hyperbon: :hyperbon: :hyperbon:  :D

modesti

  • CàA
  • Messages: 17870
  • Boinc'eur devant l'éternel
  • *****
  •   
    • Brocantes en Bourbonnais
Re : Re : Formula BOINC 2020
« Réponse #9 le: 08 octobre 2020 à 10:09 »
@dudumomo: I don't know if the project admins are contacted ahead of a sprint. I’m just “playing the game” not taking part in the administration ;)

I made some archaeology in other sections of this forum regarding FB.
There is/was a steering committee made of 9 members of 5 different teams. In fact, most of its members haven’t shown up since early 2019 2013. I suppose that’s why the admin of the FB homepage put up the votes for the marathon projects and the sprints. You can say that he’s alone to keep things up and he was about to throw in the towel last year several times.

When you look at the number of votes for a sprint (e.g. the latest one http://formula-boinc.org/admin/sprint.py?id=18&action=display) you can see that only 10 out of 77 teams voted. That’s not very encouraging. Either most of the teams aren’t interested in FB anymore, or they don’t know about the vote (which could be a result of the lack of interest). It would be interesting to see how many teams of each division showed up for the vote.

@Icecold: Maybe you can offer your services to Sébastien, the FB admin, to help him get things done? I think he definitely needs some helping hands at least for the contacts with the projects and maybe with the registered teams. Like most of us he’s busy IRL.

Demande aux modos: Serait-il possible de séparer la discussion à partir d’ici https://forum.boinc-af.org/index.php/topic,8166.msg505862.html#msg505862 et de créer un nouveau topic “Discussions on Formula Boinc [in English]”. Merci d’avance!
(Attention, il y a un intrus :D l’annonce du prochain sprint)

Edit: correcting dates
« Modifié: 08 octobre 2020 à 10:13 par modesti »

Viendez chez nous, cause qu'on est les meilleur(e)s :D


In memoriam Jip

ousermaatre

  • Gentil admin
  • Messages: 11944
  • Boinc'eur devant l'éternel
  • *******
  •   
Re : Discussions on Formula Boinc [in English]
« Réponse #10 le: 08 octobre 2020 à 14:46 »
c'est fait.

modesti

  • CàA
  • Messages: 17870
  • Boinc'eur devant l'éternel
  • *****
  •   
    • Brocantes en Bourbonnais
Re : Discussions on Formula Boinc [in English]
« Réponse #11 le: 08 octobre 2020 à 15:40 »
Merci grand pharaon :agenoux:

Viendez chez nous, cause qu'on est les meilleur(e)s :D


In memoriam Jip