Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
![]() |
World Community Grid Forums
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
No member browsing this thread |
Thread Status: Active Total posts in this thread: 20
|
![]() |
Author |
|
TBirdTheYuri
Advanced Cruncher France Joined: Mar 5, 2006 Post Count: 115 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
J'ai un petit souci avec quelques unités DDDT2. Les estimations initiales de temps de calcul étaient erronées et je me retrouve avec 5 unités qui ne seront pas finies à temps (jusqu'à 40 heures de retard pour la plus en retard).
![]() Est-ce que ces unités seront malgré tout considérées comme conformes, et donc créditables, ou bien seront-elles considérées comme nulles et donc en statut "trop tard" sans crédit? ![]() En gros, est-ce que je dois garder ces unités dans mon cache ou bien est-ce qu'il est préférable de les abandonner dès maintenant? ![]() Les unités qui suivent cette série de 5 DDDT2 ne posent elles pas de problème, la deadline est suffisamment éloignée. ![]() --- I have a small problem with some units DDDT2. The initial estimates of computing time were misguided and I find myself with 5 units that will not be finished on time (up to 40 hours of delay for more late). ![]() Are these units still will be considered compliant, and therefore credit, or will they be regarded as null and therefore status "too late" without credit? ![]() Basically, do I have to keep these units in my cache or is it is better to quit now? ![]() The units that follow this series of 5 DDDT2 they pose no problem, the deadline is so remote. ![]() |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
If the results are too late, they will not be used and no credit is given.
The day after the deadline the work is released so that someone else gan give it a try. Please do the good thing and cancel the ones you cannot return on time for sure. |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
If the results are too late, they will not be used and no credit is given. As far as I know, this is not correct... When results are overdue they normally get aborted by the project (server aborted). If they already have begun, before the project aborts the WU, the credit will be granted, as long as you return the result before the repair job arrives. And if it is returned before the repair job starts, the repair job will be aborted by the project. |
||
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3715 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
TBird,
----------------------------------------Although you have given little info I have the feeling that these WUs might be type A or at least type B WUs. Since these WUs are conditioning the rest of the process, the sooner you abort them, the better. That will allow immediate redistribution to a faster returner. Maybe you should consider reducing your cache size? That way you would be less exposed in case the DCF goes off track. ________________________ Bien que tu aies donné peu d'information j'ai le sentiment qu'il pourrait s'agir de tâches de type A, ou au moins de type B. Comme ces tâches conditionnent la suite du process, les abandonner au plus tôt permettrait de les redistribuer immédiatement à un cruncher rapide. Peut-être que tu devrais réduire la taille de ton cache? De cette façon tu serais moins exposé lorsque le DCF "dérape". |
||
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3715 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
The day after the deadline the work is released so that someone else gan give it a try. Arnolddepater,When the deadline is passed the replacement WU is distributed within seconds. |
||
|
X-Files 27
Senior Cruncher Canada Joined: May 21, 2007 Post Count: 391 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Maybe you should consider reducing your cache size? That way you would be less exposed in case the DCF goes off track. I think a lot of crunchers forgot that the deadline is 8 days and their cache size is set to 10. ![]() ![]() ![]() |
||
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3715 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I think a lot of crunchers forgot that the deadline is 8 days and their cache size is set to 10. Now that the return delay of A-type WUs has been set to 8 days (reliable distribution mode) these big cache machines should no longer see any of these (requested average return delay < 2 days). But with the overestimated C-type WUs of the first day it is possible that a few 2-day caches have been filled with 10 days of work or even more on that day.![]() |
||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
The day after the deadline the work is released so that someone else gan give it a try. Arnolddepater, When the deadline is passed the replacement WU is distributed within seconds. It's not that straight from what I'm keyhole observing: - See quite a few repairs that actually go out about 4-5 minutes before the expiration of the No Reply result. Don't know why, but there being maybe a speculation of getting the No Reply to abort, maybe the overdue client forced to talk to the servers to find out if it should be started at all ????????? - Some repair tasks take an hour and more to find the right wingman of the correct reliability class. Posted such an item in the HFCC forum some hours ago [could be paused validators too]. Generally though many will go out within minutes, depending on demand of the moment. Je pense donc... edit: quoted one post and the next one got plugged in :shrug:
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 2 times, last edit by Sekerob at Mar 30, 2010 4:08:01 PM] |
||
|
TBirdTheYuri
Advanced Cruncher France Joined: Mar 5, 2006 Post Count: 115 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
JmBoullier : Ce sont effectivement des type A à très longue période. Elles sont pour deux d'entre elles déjà entamées, les trois autres le seront dans la nuit.
----------------------------------------Si elles ne seront pas considérées comme nulles, alors autant que je les finisse. Je laisse un cache relativement important pour m'éviter un suivi manuel trop répétitif, le projet étant souvent à sec. Habituellement ça passe large, juste que là il a pris ces quelques unités en estimant 25 heures alors qu'il leur en faut pas loin du double. Celles qui suivent ne posent pas de souci. --- JmBoullier: This is actually a type of very long period. They are two of them already opened, the three others will be in the night. If they will not be considered as null, then it seems that I finish. I left a relatively large cache to avoid me a manual tracking too repetitive, the project is often dry. Usually it goes off, just as there has been few units by estimating 25 hours while they require no further twice. Those that follow are not a concern. [Edit 1 times, last edit by TBirdTheYuri at Mar 30, 2010 7:52:28 PM] |
||
|
boulmontjj
Senior Cruncher France Joined: Nov 17, 2004 Post Count: 317 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Bonjour,
----------------------------------------J'ai également un problème avec quelques WU que j'ai reçues le 24 mars vers 10h30. J'en ai reçu 27 au total. Quelques WU sont à retourner pour le 8 avril et là, pas de problème par contre, les autres sont à rendre pour le 1er soit aujourd'hui. Et là, j'en ai 6 qui ne passeront pas plus 4 qui sont en cours et que je vais tout de même tenter de laisser tourner jusqu'au bout car je n'ai pas envie de perdre le travail qui a été fait sur ces unités. Par contre, je tenais à dire que je trouve que des unités demandant près de 50 heures de calcul devraient avoir une date de retour plus longue que des unités normales car 8 jours, cela peut faire court pour certain membre, et vous risquez de vous retrouver avec une avalanche de "pas de réponse". |
||
|
|
![]() |