Uploading
Message boards :
Number crunching :
Uploading
Message board moderation
Previous · 1 . . . 6 · 7 · 8 · 9 · 10 · 11 · 12 . . . 15 · Next
| Author | Message |
|---|---|
 Jim Baize Send message Joined: 6 May 00 Posts: 758 Credit: 149,536 RAC: 0 
|
So, if I understand you correctly, you want to throttle down the WU's leaving Berkeley, basically reducing the number of WU's going out. That is a good thought. With my limited knowlege I don't see why it wouldn't work in theory. Jim PS I am assuming FFEMTc is FireFighter Emergency Med Tech... but what is the "cJ"? I have a thought. Since the servers are currently at 100% capacity (my understanding... I don't understand it all so tell me if this cant work)... Make it to where 1 out of every 6 hours, or however many they want, the servers only accept uploads to try to get things caught up. So 4 times a day, 4 hours a day, no one will be able to download (which if everyone has it set up right, they have a couple of extras sitting on their computer anyway) and everyone can upload. Then when things get caught up, change it to three total hours a day, then two.. and find out where things start getting backed up again and leave it there.. The downloading isn't the important thing, because people can store extra's on their computer.. So an hour at a time not being able to download shouldn't hurt anyone. ID: 137527 ·  |
Qui-Gon  Send message Joined: 15 May 99 Posts: 2940 Credit: 19,199,902 RAC: 11
 |
I can't upload the completed WUs. Any suggestions? But interesting reading, isn't it? It runs the gamut . . . ID: 137528 · |
Misfit  Send message Joined: 21 Jun 01 Posts: 21804 Credit: 2,815,091 RAC: 0
|
I can't upload the completed WUs. Any suggestions? On his printer? Read from the monitor and save some trees! me@rescam.org ID: 137529 · |
Jord  Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 
|
Concern about what exactly? Credits? RAC? Quota per Day? Nothing bad will happen to the Project, nor to the clients. If at that moment you mean the Boinc Core Clients. They will just try to upload the units until well after the deadline of those units, whether they are met or not. All that'll happen to people's credits is that they won't see any, their RACs will drop and their quota will drop as well. Hopefully by that time, another outage will have changed the code that now says if you have one unit over deadline, that you will start from 1 unit a day in your quota, only to increase over the time you upload & report more units correctly. Just don't patronise me, mr. Lowfield. Why didn't you follow your initial thoughts in your thread Bye Bye Boinc? Because at that time you didn't understand how things around here worked. Has that changed so much that you think you know how everything works? Or has the whole thread done nothing to you? ID: 137531 · |
|
EclipseHA Send message Joined: 28 Jul 99 Posts: 1018 Credit: 530,719 RAC: 0
|
I agree with Tigher. End Classic Now! The classic folks have been told that there will be a "global email" to all users prior to the shutdown of Classic. On top of that, IIHO, the shutdown of classic should be done in an orderly (and pre-documented - the last "plan is way outdated)) way. The truth is there might still be more people (and note I didn't say hosts) crunching classic than are crunching seti/boinc. Hosts don't care when they can't get data, but I think it's safe to say, people do! If they shutdown classic on monday to reuse the HW for seti/boinc, trust me, it would anger a whole bunch of people that are donating resources (CPU time and their time) (A.K.A. "Volinteers"), and would not be a wise move. I think the solution (as we all know classic is going away), is for UCB SSL to publish a "transition plan" (with real dates), here and on the classic site. It's got to be a "reasonable timeframe" for the events. For example, "Send email mon, stop work tue, stop acepting work wed, reuse boxes on thur" would not be good! What it seems some don't understand is the real "volinteers" here are not the handfull of folks that write code or document the program. It's the thousands, or more correctly hundreds of thousands that crunch the data. Without the real volinteers, the folks at UCB SSL would sit at a monitor and say "hey, we got a result.. That's 5 today and almost 30 this month!" ID: 137532 · |
Astro  Send message Joined: 16 Apr 02 Posts: 8026 Credit: 600,015 RAC: 0 |
I have a thought. Since the servers are currently at 100% capacity (my understanding... I don't understand it all so tell me if this cant work)... Make it to where 1 out of every 6 hours, or however many they want, the servers only accept uploads to try to get things caught up. So 4 times a day, 4 hours a day, no one will be able to download (which if everyone has it set up right, they have a couple of extras sitting on their computer anyway) and everyone can upload. Then when things get caught up, change it to three total hours a day, then two.. and find out where things start getting backed up again and leave it there.. The downloading isn't the important thing, because people can store extra's on their computer.. So an hour at a time not being able to download shouldn't hurt anyone. This Idea has merit as long as those of us on dial up don't have to wait for an hour and a half to get and send WU's during times of normal operation. It takes me about 1/2 hour just to download the wus. So, If a user isn't home at just the right time it would certainly be discouraging. tony ID: 137533 · |
|
chrisjohnston Send message Joined: 31 Aug 99 Posts: 385 Credit: 91,410 RAC: 0
|
So, if I understand you correctly, you want to throttle down the WU's leaving Berkeley, basically reducing the number of WU's going out. Initials... But basically.. Just 4 hours a day... No one can download.. The rest of the day the downloads and uploads are both available... But during those four hours, the CPU cycles spent on sending files are not used for sending but recieving. It sounds like all they would have to do is write a script for the program.. Way above me, but for someone who knows what they are doing, shouldn't be too bad. - cJ  
ID: 137535 · |
|
Jim Baize Send message Joined: 6 May 00 Posts: 758 Credit: 149,536 RAC: 0
|
Ok... so lets modify the plan a little. instead of 4 x 1 hour periods how about 8 x 0.5 hour periods or maybe even 16 x 0.25 hour periods? I think it would have an effect similar to what Chris was wanting while still keeping the people on dial-up from having to wait. Shoot... even say the first 5 minutes of every hour dedicated to the servers recieving only. many little bitty steps add up over time. Jim This Idea has merit as long as those of us on dial up don't have to wait for an hour and a half to get and send WU's during times of normal operation. It takes me about 1/2 hour just to download the wus. So, If a user isn't home at just the right time it would certainly be discouraging. ID: 137536 · |
|
chrisjohnston Send message Joined: 31 Aug 99 Posts: 385 Credit: 91,410 RAC: 0
|
I can't upload the completed WUs. Any suggestions? Didn't print it.. Just been following it and know that there is alot of stuff, so I hit print preview. - cJ
ID: 137537 · |
|
EclipseHA Send message Joined: 28 Jul 99 Posts: 1018 Credit: 530,719 RAC: 0
|
I have a thought. Since the servers are currently at 100% capacity (my understanding... I don't understand it all so tell me if this cant work)... Make it to where 1 out of every 6 hours, or however many they want, the servers only accept uploads to try to get things caught up. So 4 times a day, 4 hours a day, no one will be able to download (which if everyone has it set up right, they have a couple of extras sitting on their computer anyway) and everyone can upload. Then when things get caught up, change it to three total hours a day, then two.. and find out where things start getting backed up again and leave it there.. The downloading isn't the important thing, because people can store extra's on their computer.. So an hour at a time not being able to download shouldn't hurt anyone. This solution would take far longer to implement and get communicated than it "should" take for the servers to recover. The problem now is that it's the weekend and I'll bet there will be nothing happening until UCB wakes up on Monday. That's been the story for years. ID: 137539 · |
|
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3
|
I wouldn't go for a 5 minute upload window. That would just put loads of people's PCs into upload queue's again. Hour by hour would do fine, I think, since I don't see any computer out there that does Seti units within the hour, outside of the noisy units. ID: 137541 · |
|
chrisjohnston Send message Joined: 31 Aug 99 Posts: 385 Credit: 91,410 RAC: 0
|
I agree with Tigher. End Classic Now! Timeline... eMail sent out on Wednesday of this week. September 1, no more downloads. September 15, server transfered to BOINC as client upload (UCB download). October 1, no more uploads. October 15 server transfered to BONIC as client upload. October 16, all the VOLUNTEERS (note correct spelling) are happy. Things might still not be 100% due to the added BONIC users, however, atleast all of the resources will be dedicated to BOINC, which is obviously where the project is going. Problem solved IMHO. Chris - cJ
ID: 137542 · |
|
Jim Baize Send message Joined: 6 May 00 Posts: 758 Credit: 149,536 RAC: 0
|
I think it would be interesting to put both schemes to the test to see if one out perfomed the other while keeping the dial-up users in the loop. Jim I wouldn't go for a 5 minute upload window. That would just put loads of people's PCs into upload queue's again. Hour by hour would do fine, I think, since I don't see any computer out there that does Seti units within the hour, outside of the noisy units. ID: 137546 · |
|
David Stites Send message Joined: 22 Jul 99 Posts: 286 Credit: 10,113,361 RAC: 0
|
?[/quote] Who would've thought that my little comment 3 days ago would have sparked a 58 page (on my printer) conversation/debate/arguement/discussion/insertyourwordhere about the situation.[/quote] But interesting reading, isn't it? It runs the gamut . . .[/quote] On his printer? Read from the monitor and save some trees![/quote] This reminds me of years ago when I was a electrician. One of my least favorite jobs was pulling wire, especially the big wire pulls. 3 or 4 men would set it up and then need 5 to 8 more guys to help. Without fail one or more of the *helpers* would try to take over and reorganize the setup. Once when this happened I told the wannabe alpha male that we didn't need any help replanning, that we had it covered. When he wouldn't give up telling us what to do with our job I asked him what his foreman had told him to do. He looked surprised and said "To come over and help us with the pull." I then said that we wanted him to go over by a paticular reel of wire and make sure it didn't tangle up as it was coming off. And if he couldn't do that to go back to his foreman and tell him that he couldn't help us. That shut him up and a couple of hours later the wire was in the conduit. I was asked to help, not take over the job. I've been doing that for a while, and there have been some problems, but I think an amazing amount of work has been done and I am glad I helped. I will continue to help as long as it is wanted. SETIATHOME was only supposed to last a couple of years but it has evolved into BOINC and is a substanstial achievement in advancing scientific knowledge. Congrats to SETI, David Stites Pullman, WA USA 
ID: 137550 · |
Webmaster Yoda  Send message Joined: 3 Apr 99 Posts: 52 Credit: 500,125 RAC: 0 
|
Perhaps I've been lucky, or I've stumbled on a workaround... After having some WUs sitting in the queue for hours, I decided to take a chance and aborted the transfer. At worst I'd lose 1.5 hours worth of crunching. To my surprise, after I aborted the transfer, the WU uploaded almost immediately and it's now showing as complete, with credits pending. Maybe it was a fluke, but perhaps it's an answer to our problems (and maybe it's a clue to the seti sys admins too) *** Join the #1 Aussie Alliance on SETI *** ID: 137551 · |
|
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3
|
So if a dial-up user has missed the last 5 upload windows on his AMD 3800+, he has some 200 units waiting to upload in 5 minutes? It takes every unit seperately about 10 seconds to get an answer from the scheduler, 200 units at average 15KB on a 4.5KB upload won't go fast. Around 670 seconds. Think you can zip them up? ;) ID: 137552 · |
|
chrisjohnston Send message Joined: 31 Aug 99 Posts: 385 Credit: 91,410 RAC: 0
|
I have a thought. Since the servers are currently at 100% capacity (my understanding... I don't understand it all so tell me if this cant work)... Make it to where 1 out of every 6 hours, or however many they want, the servers only accept uploads to try to get things caught up. So 4 times a day, 4 hours a day, no one will be able to download (which if everyone has it set up right, they have a couple of extras sitting on their computer anyway) and everyone can upload. Then when things get caught up, change it to three total hours a day, then two.. and find out where things start getting backed up again and leave it there.. The downloading isn't the important thing, because people can store extra's on their computer.. So an hour at a time not being able to download shouldn't hurt anyone. "Communicated": Post Monday morning on the home page and technical news page that it will be implemented. "Implemented": Shouldn't take more than a couple of days at most to make this happen for someone who knows what they are doing. "Should Take": Can you define this for me, because so far it has been almost a week. How long is "Should Take"? And, as the total number of people and computers doing the work, the "Should Take" will keep expanding due to greater demand. Therefore, "Should Take" can not be defined. So, "Should Take" can be an hour, a day, a week, a month, a year. And each time there is an outage, the "Should Take" will get longer, and longer and longer. So why not atleast throw something in there (my idea, or yours if you come up with one that is better) that will atleast alleviate part of the problem..... Or atleast make it APPEAR as though they are trying to take a step in the right direction by doing something to attempt to fix the problem. - cJ
ID: 137553 · |
Fuzzy Hollynoodles  Send message Joined: 3 Apr 99 Posts: 9659 Credit: 251,998 RAC: 0 |
Hear hear!!! And with BOINC it has become a landmark in distributed computer technology!!! And with this we are able to join all the wonderful projects out there! "I'm trying to maintain a shred of dignity in this world." - Me 
ID: 137554 · |
|
Jim Baize Send message Joined: 6 May 00 Posts: 758 Credit: 149,536 RAC: 0
|
I read about this "work around" already today. Although I've personally not tried it, several others were saying that it really doesn't upload it, but rather kills the WU and the result was showing as invalid. Just something to keep in mind before running headlong into this work around. Jim Perhaps I've been lucky, or I've stumbled on a workaround... ID: 137555 · |
John McLeod VII  Send message Joined: 15 Jul 99 Posts: 24806 Credit: 790,712 RAC: 0
|
It was expected that for the first 2 - 3 days, there would be trouble uploading. However, after that the problems became alarming and got the attention of the developers. We shall have to see how what they find, and how it can be resolved. There is nothing further anyone can do without looking at the code and the hardware.   BOINC WIKI ID: 137556 · |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.