Message boards :
Number crunching :
Now enough about SETI. I leave
Message board moderation
Previous · 1 · 2 · 3 · 4 · Next
Author | Message |
---|---|
Rick Send message Joined: 27 Mar 01 Posts: 8 Credit: 25,223,770 RAC: 0 |
SETI could have backup servers in place to carry on until the main servers are ready to start again that is a way some commercial servers work such as at dynDNS they can supply back up servers until the SETI servers get back on line. dynDNS offered the service to me but since I am not a commercial server I host at my home computer I do not need their back up servers SETI could do something similar. I feel your pain Philippe. Yes but organziations such as dynDNS are commerically supported organizations. SETI, being donation supported, doesn't have the resources to do this. Unless someone were to donate some serious hardware, that is. |
Paul D Harris Send message Joined: 1 Dec 99 Posts: 1122 Credit: 33,600,005 RAC: 0 |
SETI and BOINC uses peoples home computers for its clients. Why can't they use people home computer instead of just clients why not as servers and clients or just servers as a computed server network. Bye Paul |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
SETI and BOINC uses peoples home computers for its clients. Why can't they use people home computer instead of just clients why not as servers and clients or just servers as a computed server network. ... because the data has to start at the telescope, then get to the download server and scheduling server, and then upload to the upload server, and go to the science database (assimilation and transition). When you try to "distribute" the server side, all you've really done is move the problem around -- the data still has to get in and out of Berkeley, it doesn't matter if it goes from Berkeley to a "distributed server" to a cruncher, or from Berkeley straight to a cruncher. To put it another way, in a sense BOINC already does this. If you download 100 work units, and report (and download) 20/day, you've always got 4 days worth of work -- you don't need to talk to Berkeley for 3 or 4 days. This is why I've been saying that this whole "Panic mode -- servers down" worry is basically crazy. If you have work, and you have pending uploads, the servers will be back, and uploads will happen when they're back. 99.999% reliability just is not needed. 80% is probably enough. That's demonstrated every Tuesday, but we still don't seem to get it. |
Gary Charpentier Send message Joined: 25 Dec 00 Posts: 30640 Credit: 53,134,872 RAC: 32 |
SETI and BOINC uses peoples home computers for its clients. Why can't they use people home computer instead of just clients why not as servers and clients or just servers as a computed server network. Tuesday's only show about a 97% up time, about 4 hours a week. Getting down to 80% is a lot more off time, more than 24 hours a week. I'm sure the project would be fine with that, but the RAC chasers would have heart attacks. |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
SETI and BOINC uses peoples home computers for its clients. Why can't they use people home computer instead of just clients why not as servers and clients or just servers as a computed server network. ... and my point is that the project, at somewhere around 97% is well above the reasonable minimum reliability needed. This is not Amazon.com. People are not inconvenienced because the site is down, it's only a question of perception, nothing to do with reality. |
Paul D Harris Send message Joined: 1 Dec 99 Posts: 1122 Credit: 33,600,005 RAC: 0 |
If I had an organization would I want a good professional reputation as being reliable or would I want a reputation as being down and an inconvenience to the people who use it which only turns away clients instead of encouraging participation. After all Berkley does has a good reputation it’s SETI servers should be likewise. |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
If I had an organization would I want a good professional reputation as being reliable or would I want a reputation as being down and an inconvenience to the people who use it which only turns away clients instead of encouraging participation. After all Berkley does has a good reputation it’s SETI servers should be likewise. It really helps to read some of the design documents on the BOINC website. This one is a good overview. ... and I would argue that we're applying the wrong standard, because we're looking at this wrong. In the document I referenced, in section 2.1 it says "reduce the barriers of entry to public resource computing" and therein lies the rub. For SETI to have the kind of reliability many of the vocal users demand, they would have to spend dramatically more money: double the number of servers, redundant connections, maybe space in a data center someplace -- and more staff! It's very expensive to get that kind of reliability. ... and if SETI has to spend serious money to get to five-nines, then BOINC is a complete failure. It means you can't run a big volunteer computing project on a shoestring budget. If on the other hand, your computer always has work to crunch, even if the BOINC servers are down for maintenance, weekly outages, then the BOINC system (Client and Server components), and as long as that work gets reported and credit granted (absent failures on your computer, of course) then BOINC is a resounding success. Is SETI pushing the envelope, given the number of participants, under-sized hardware, and the vanishingly small budget? Absolutely, and more. But instead of panic every time there is a momentary glitch, we should be celebrating the resilience in the overall BOINC system. ... and we shouldn't care about server reliability if the BOINC client deals with it automatically. |
marsinph Send message Joined: 7 Apr 01 Posts: 172 Credit: 23,823,824 RAC: 0 |
Hello everybody. I have read all the post. I conclude some tryes to recover me to work for other project ! Wrong. You have not read the first post ! Some ask me to connect at some defined hours. Also please read the first post. Some discuss between their own experience and answer others, but i think it is not the goal of this post ! Some propose to increase the cache. But it not solve the problem because my cache is already very large ( about 6.000.000) but server have also deadline for wu. When I receive 200WU with deadline after 5 days. No problem. Computer do it. But is not able to send due to server problems!!! OzzFan, i will remember your previous post. Thanks. |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
Hello everybody. Again, very hard to help because (apparently) we're having trouble communicating. I read your earlier post, and it sounds like you can't connect anytime at all, now you can't connect on a schedule either. I'm confused. If the servers are down, deadlines "extend" because of they way deadlines are processed. |
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
I conclude some tryes to recover me to work for other project ! Wrong. You have not read the first post ! I have to disagree. Your first post did state you would work for another project who is more stable: GOOD BYE SETI . I will go to a other project who is more stable |
H Elzinga Send message Joined: 20 Aug 99 Posts: 125 Credit: 8,277,116 RAC: 0 |
Hello everybody. Looks a bit like we get the blame for not solving his company internal problems. The frequently referd first post is very unclear but should explain to us a apperantly very complex problem he has at his workplace. |
rebest Send message Joined: 16 Apr 00 Posts: 1296 Credit: 45,357,093 RAC: 0 |
Hello everybody. I certainly did read your first post. I also read your statement: The firewall is configured to allow connection to Seti at defined hours. But when it is not able to connect to Seti, the next "open port" will be one week later. If you were to configure your firewall to allow connection shortly after the regularly scheduled weekly server maintenance at Berkeley each Tuesday, your connections would be much closer to 100%. If the "goal of this post" is just to complain that the servers are not available 100% of the time when you need them on weekends and so you are discontinuing the project, that's OK. I understand. Under the current limitations, the servers will never be available 100% of the time and the potential for extended outages is greatest during the weekend. For most of us, we have found ways to work around this problem and we have tried to share them with you. I'm sorry that they haven't helped. Good luck. Join the PACK! |
Dandelion Send message Joined: 15 Sep 02 Posts: 1 Credit: 1,000,064 RAC: 0 |
Philippe, i don't know if you follow this thread any further. Anyway... Please, read carefully. I need to open firewall manually on each computer due to restriction on logical port due to firewall security at office (Ministry of....of Belgium. To help you with complex configurations you have to supply us with more information about your "office network". Maybe this have to be secret (to us). Company networks usually have an internal network (intranet). The central it-department defines the rules how it must be used. The connection from the intranet to the outer world (internet) is centralized and usually protected thru firewall, virus scanner and proxy server. All network communication from intranet clients with the internet has to use the proxy server (there can be exceptions); NO DIRECT communication from intranet clients to the internet. So the proxy has to be used with boinc. You can place an own proxy with firewall between the "company" proxy and your boinc clients. So there is a central place where you can manage how and when boinc clients can communicate with the internet. But the main problem may be the given rules in your ministry, which we dont' know. Regards Dandelion |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
Yes, we do know. Phillipe said earlier it wasn't that he didn't know how to set it up so he didn't have to babysit the rigs, but that policy limited his flexibility in what he could do with access. Whether that policy makes any sense in the context of BOINC and any additional security risk it represents by allowing more flexibility is moot. They are not his machines and network to do with what he wants. His main complaint was that every time the connection opportunity window opens SAH seems to be 'down', so it's game over for those hosts a lot of times. Alinator |
marsinph Send message Joined: 7 Apr 01 Posts: 172 Credit: 23,823,824 RAC: 0 |
Yes, my computer at home !!! But then it is the same ! I will not spend all my time to copy wu then UL/DL when seti is not stable. So all the wu are out of date. Because the close of port 80 by seti, i already lost about 2000WU ! |
The Gas Giant Send message Joined: 22 Nov 01 Posts: 1904 Credit: 2,646,654 RAC: 0 |
Hello everybody. I would have thought that if you can contact the server to get work then at the same time you can upload work and report it. Go figure... |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
They are different physical servers, so maybe, maybe not. |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
Because the close of port 80 by seti, i already lost about 2000WU ! Okay, now I'm confused. I don't see anything in BOINC that moves off of the standard HTTP port when communicating between the project and users. As far as I know, the only thing that doesn't use a standard port is the communication between BOINC and the BOINC Manager, and that's kind-of optional. ... and I vaguely remember, you can override that if you wish. |
petros Send message Joined: 10 Jul 03 Posts: 72 Credit: 141,587 RAC: 0 |
Yes, my computer at home !!! i think you have to reconfigure your firewall, something makes me believe that your firewall probably blocking the upload. Perhaps it will help if you tell us what operating system you have and what type of antivirus and firewall are installed on your pc. SETI |
OzzFan Send message Joined: 9 Apr 02 Posts: 15691 Credit: 84,761,841 RAC: 28 |
Its not that SETI doesn't use port 80, its that it also uses port 443 and he doesn't like that. |
©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.