| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: Is WCG upping the grid efficiency just a tiny little more? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 4
|
|
| Author |
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Only today made the first for me observatin that there's no waiting on Who's first to return the second copy after the Make-Up job has been sent? Now the credo seems to have advanced to: You had X days... did not do it, someone else will, and should, for a server abort goes towards the daily quota, so the volunteer who is a fast returner should not suffer the consequence when the Too Late comes in after deadline. The tolerance still might be 24+ hours from reading the distribution flow (or the Too late never communicated with the servers in the intermediate?)
----------------------------------------Project Name: Help Conquer Cancer Created: 27-2-10 Name: X0000084380440200704051431 Minimum Quorum: 2 Replication: 2 Result Name App Version Number Status Sent Time Time Due / Return Time CPU Time (hours) Claimed/ Granted BOINC Credit X0000084380440200704051431_ 2-- - In Progress 11-3-10 06:27:26 15-3-10 06:27:26 0.00 0.0 / 0.0 < my client about finishing this. X0000084380440200704051431_ 1-- 606 Pending Validation 1-3-10 06:34:53 2-3-10 17:09:52 8.54 68.5 / 0.0 X0000084380440200704051431_ 0-- 606 Server Aborted 1-3-10 06:31:31 12-3-10 12:25:45 0.00 0.0 / 0.0 < former no reply Maybe there's in 6.10.36 a signaling to the server: "reliable host crunching it now" coming? Here though the No Reply got sent off without waiting. Goodos to the WCG techs... no point of getting 3 copies when 2 is well and truly scientifically sufficient. (Yes, Yes, I've got my own pace of discovering things ;>)
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Ingleside
Veteran Cruncher Norway Joined: Nov 19, 2005 Post Count: 974 Status: Offline Project Badges: 
|
(Yes, Yes, I've got my own pace of discovering things ;>) Well, the particular server-aborting-code being used in this example haven't been changed since it was added to BOINC in April 2007, so nothing has changed performance-wise. The reason for not seeing this before is it's likely fairly seldom someone exceeds their deadline on tasks that haven't started yet, but granted there likely is a few having extended computer-problems or is having a holiday or something... If the user had been running v6.6.xx or later client, the task wouldn't show-up as server-abort, but instead would have been aborted by client instead for not starting by the deadline.  Maybe there's in 6.10.36 a signaling to the server: "reliable host crunching it now" coming? The client haven't any idea where's anything called a "reliable computer", and I don't remember any plans of adding any reporting by client when passing deadline. But, the client does tell the application about the deadline, meaning the application can example exit early in case can't meet the deadline, and can upload the current checkpoint-file so can be issued to someone else that finish the task or something. AFAIK no projects has currently added any code to any of their applications to exit early or anything...  "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Think you thoroughly missed the difference from before. It's not important to me that you did. Those that do see it too know what it is.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Ingleside
Veteran Cruncher Norway Joined: Nov 19, 2005 Post Count: 974 Status: Offline Project Badges:
|
Think you thoroughly missed the difference from before. It's not important to me that you did. Those that do see it too know what it is.  Well, maybe I've overlooking something, since it's not so easy to read the badly-formatted info about task-status, but the only unusual I'm seeing is that the 3rd. copy was maybe created before the 1st. tasks deadline, but since WCG doesn't display the actual deadline on this task I can't say this for sure... If you're aiming at the "waiting on Who's first to return", becoming "server abort" instead, this isn't so common but isn't anything new... "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." |
||
|
|
|