Damn You People Using Boinc 4.13

Message boards : Number crunching : Damn You People Using Boinc 4.13
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · Next

AuthorMessage
Profile Jim Baize
Volunteer tester

Send message
Joined: 6 May 00
Posts: 758
Credit: 149,536
RAC: 0
United States
Message 138392 - Posted: 18 Jul 2005, 18:13:37 UTC - in response to Message 138384.  
Last modified: 18 Jul 2005, 18:13:59 UTC

If memory serves me correctly when they had this problem it affected more than just the occasional WU, therefore it wouldn't even get to the point of being uploaded by anyone to even get validated.

Jim

If so it's a real pity.

I can't see why it would cost them too much to differentiate the network induced errors (timeout etc) from the rest (improper signing?).

It must be more expensive to drop a scientifically correct wu that has been validated ok by one or two others!


ID: 138392 · Report as offensive
Profile Kevin Beasley

Send message
Joined: 27 Aug 99
Posts: 15
Credit: 5,059,412
RAC: 16
United Kingdom
Message 138447 - Posted: 18 Jul 2005, 19:32:18 UTC - in response to Message 138098.  

It may well be like myself (a few weeks ago) that they do not know that there is a "better" version available to them.
Please do not "damn" people out of ignorance.
They are not all dedicated crunchers like you & I!

Regards,
Graeme Murphy.
www.setiuk.com


Better version than 4.13? Holy cow where have you been? The last of the "best" versions was 4.19 and at that point in time the BOINC devs should have asked the projects to alter the min required release to 4.19 to overcome the severe problems with the consequences of 4.13 not being able to fully download a file. Well at least 2 months later.

The date stamp on the BOINC V4.13 on my system shows 29 Oct 2004. The problems with 4.13 were quickly highlighted on Predictor and we went through a few versions before 4.19 was found to be a very stable release (date stamp 27 Jan 05). This was the recommended release (and still is on some projects) for quite a while until 4.25 and the BOINC Manager came along (date stamp 21 March 05). Recently V4.45 (date stamp 10 June 05) became the recommended version. I think there has been sufficient time for people to have upgraded from 4.13 to at least 4.19 by now.

If folks don't have 4.19 and want it all they need to do is ask the question and a heap of people can point them in the right direction to find it.

DAMN YOU PEOPLE USING BOINC 4.13.

I'm lucky enough not to have been adversely affected just yet but this host http://setiathome.berkeley.edu/results.php?hostid=1184839 came pretty close. Check out the wu's and notice that each wu has at least 1 download error and each of those is due to a host running 4.13!

I can't wait for the major version change to take place, but then beta is only at 4.71 so we have a while to go.

Live long and crunch.


While I agree in principle that people shouldn't use older versions of S@H, if only to ensure accuracy of compared results, I must ask the following questions:

Where were the notifications that newer versions are available?

Where are the big banners on the home page?

I even remember a time when we used to get emails telling us when new versions were released.

Not everyone has the time or inclination to check for new versions, especially those people on dial-up, and when you consider that the latest version is 6.34Mb, for those people on dialup that can be a chunk of time and money.

I also find myself questioning how the updates are handled. According to my BOINC Manager, I'm running version 4.25. However, according to the downloads page at http://boinc.berkeley.edu/download.php, the latest version is 4.45 and the previous version is 4.19 - would anyone care to comment??????

(And before you ask, YES, I'll be upgrading in the next few minutes.)

Now, back to finding ET.......................

ID: 138447 · Report as offensive
Bill Barto

Send message
Joined: 28 Jun 99
Posts: 864
Credit: 58,712,313
RAC: 91
United States
Message 138515 - Posted: 18 Jul 2005, 20:58:29 UTC - in response to Message 138384.  

If so it's a real pity.

I can't see why it would cost them too much to differentiate the network induced errors (timeout etc) from the rest (improper signing?).

It must be more expensive to drop a scientifically correct wu that has been validated ok by one or two others!


The newer versions don't have the problem with network induced errors. All people have to do is update their BOINC clients.
ID: 138515 · Report as offensive
Astro
Volunteer tester
Avatar

Send message
Joined: 16 Apr 02
Posts: 8026
Credit: 600,015
RAC: 0
Message 138535 - Posted: 18 Jul 2005, 21:33:51 UTC

Seti would sent out emails in Classic when the Application version changed. It didn't then have a Boinc Manager or Core Client. I haven't seen notes or anything else telling member to update and I think they should do something. Hopefully, when 5.01 (or some number) comes out the Major Version change will "Require" the users to upgrade since major version 4 core clients will no longer be able to get work, and then the users will ask "why can't I get work", or "why am I getting this message about "you are using major version 4 and need to upgrade to version 5 clients".

After problems with the upgrade to version 4.14, the devs furiously churned out one beta client after another, and if I remember correctly it only took them about a week to get up to 4.19. All versions prior to and including 4.19 do NOT use a GUI (graphical user interface). All versions after it DO contain the option of a GUI. They probably keep 4.19 around as an option to allow US a choice. Then Starting with 4.35 JM7 (John Mcleod 7) developed and implemented the "New Scheduler" which makes choices for us as to what to crunch next, how much work to get, When to get new work, and when to report it.

I hope this history helps some understand.

tony
ID: 138535 · Report as offensive
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 138572 - Posted: 18 Jul 2005, 22:18:16 UTC - in response to Message 138535.  
Last modified: 18 Jul 2005, 22:19:36 UTC

I hate to say it, Tony, but the Windows version at least has always had a GUI. It hasn't had the seperation of Boinc and the Boincmanager as we have it since 4.35.

But prior to that, if you started up BoincGUI.exe Boinc would start with the GUI.
Some versions came with a seperate BoincCLI.exe .. yep, the Command Line Interface version of the same program. What is now Boinc.exe
ID: 138572 · Report as offensive
Profile Kevin Beasley

Send message
Joined: 27 Aug 99
Posts: 15
Credit: 5,059,412
RAC: 16
United Kingdom
Message 138593 - Posted: 18 Jul 2005, 22:43:14 UTC - in response to Message 138535.  

Seti would sent out emails in Classic when the Application version changed. It didn't then have a Boinc Manager or Core Client. I haven't seen notes or anything else telling member to update and I think they should do something. Hopefully, when 5.01 (or some number) comes out the Major Version change will "Require" the users to upgrade since major version 4 core clients will no longer be able to get work, and then the users will ask "why can't I get work", or "why am I getting this message about "you are using major version 4 and need to upgrade to version 5 clients".

After problems with the upgrade to version 4.14, the devs furiously churned out one beta client after another, and if I remember correctly it only took them about a week to get up to 4.19. All versions prior to and including 4.19 do NOT use a GUI (graphical user interface). All versions after it DO contain the option of a GUI. They probably keep 4.19 around as an option to allow US a choice. Then Starting with 4.35 JM7 (John Mcleod 7) developed and implemented the "New Scheduler" which makes choices for us as to what to crunch next, how much work to get, When to get new work, and when to report it.

I hope this history helps some understand.

tony


Tony,
Thanks for the history lesson - most appreciated.

Okay, so here's Suggestion One - why doesn't the SETI team email everyone when there is a new version available? That would mean that most people would switch over as soon as they could. After all, they have all our email addresses.

Second Suggestion, when they release the next version, why don't they make the WUs twice the size? There's probably some sound explanation that I'm sure some kind soul can put in words of one syllable for me, but to my way of thinking, wouldn't that mean less splitting of the tapes and less loading on the servers of people uploading and downloading as they'd being doing it less frequently?

Regards.

ID: 138593 · Report as offensive
Astro
Volunteer tester
Avatar

Send message
Joined: 16 Apr 02
Posts: 8026
Credit: 600,015
RAC: 0
Message 138611 - Posted: 18 Jul 2005, 23:05:14 UTC - in response to Message 138593.  
Last modified: 18 Jul 2005, 23:05:53 UTC

Okay, so here's Suggestion One - why doesn't the SETI team email everyone when there is a new version available? That would mean that most people would switch over as soon as they could. After all, they have all our email addresses.

Second Suggestion, when they release the next version, why don't they make the WUs twice the size? There's probably some sound explanation that I'm sure some kind soul can put in words of one syllable for me, but to my way of thinking, wouldn't that mean less splitting of the tapes and less loading on the servers of people uploading and downloading as they'd being doing it less frequently?

Regards.

Both good suggestions. I know they don't have much of a staff and I don't think sending notices via email when each new version of the CC comes out is viable as they get new ones frequently, but I think they should atleast post it on the main page and/or an email when a new "Recommended" version comes out(not usually that frequently).

I don't know the science behind picking 107 seconds of data per WU, so I can't speak to that. However, the new Seti/Boinc/Beta team is testing a new Application file with doubles the sensitivity of the search and takes longer to crunch. This would slow up the network connections and may satisfy the requirements of your second Idea.

hope this helps

tony
ID: 138611 · Report as offensive
Pascal, K G
Volunteer tester
Avatar

Send message
Joined: 3 Apr 99
Posts: 2343
Credit: 150,491
RAC: 0
United States
Message 138646 - Posted: 18 Jul 2005, 23:43:55 UTC

I would love to be able to do the emails for seti, just to see how many are bad, my guess is about 20 to 30 % and I have a feeling that is low.....
Semper Eadem
So long Paul, it has been a hell of a ride.

Park your ego's, fire up the computers, Science YES, Credits No.
ID: 138646 · Report as offensive
ABT Chuck P
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 91
Credit: 316,669
RAC: 0
United States
Message 138922 - Posted: 19 Jul 2005, 6:54:10 UTC - in response to Message 138190.  

It's true that mostly old clients such as 4.13 cause excessive DL errors. But not allways. Check this result: http://setiweb.ssl.berkeley.edu/result.php?resultid=87954664


From above's last line: application version 4.18

==========================
If you look at the machines running the 4.13 CC you'll find that most are failing DL's at a rate far greater than those running 4.19 or greater. Many that I've seen attemt and fail 9 units before they get a good DL and that is done several times to fill their cache requirements.
ID: 138922 · Report as offensive
Profile AndyK
Avatar

Send message
Joined: 3 Apr 99
Posts: 280
Credit: 305,079
RAC: 0
Germany
Message 139578 - Posted: 20 Jul 2005, 13:17:33 UTC - in response to Message 138922.  

It's true that mostly old clients such as 4.13 cause excessive DL errors. But not allways. Check this result: http://setiweb.ssl.berkeley.edu/result.php?resultid=87954664


If to take a close look to that WU you see 5 DL errors with 4.13 and only 1 with 4.45.

Note: application version 4.18 has nothing to do with the core client version. It's the seti@home application which has v4.18

Andy
Want to know your pending credit?


The biggest bug is sitting 10 inch in front of the screen.
ID: 139578 · Report as offensive
Profile Ananas
Volunteer tester

Send message
Joined: 14 Dec 01
Posts: 195
Credit: 2,503,252
RAC: 0
Germany
Message 139600 - Posted: 20 Jul 2005, 14:05:37 UTC - in response to Message 139578.  

I think there's a better solution than forcing people to install different versions.

The handling of download errors is just wrong. A download error is _not_ an error result, it isn't a "result" at all. So it should not count in "Too many error results".

After each download error, the next delivery of the same work unit should be delayed but no further action should be taken.

The download errors should not even be stored in the database for long, they are usually caused by technical server problems so they are totally irrelevant for the project. Make note about the error in some log file and forget it would be more appropriate.


If there are way more DL errors with 4.13 than with 4.45 that might also mean that 4.45 isn't as popular as 4.13, if 400 try with 4.13 and 100 try with 4.45, it is only normal that there are more errors with 4.13 in such a phase with server problems. It would be interesting to count maybe over a month, how many results come in from which core version.

Otoh. I had nothing but problems with 4.13, I wonder why they do not at least upgrade to 4.19 :-/
ID: 139600 · Report as offensive
ampoliros
Volunteer tester
Avatar

Send message
Joined: 24 Sep 99
Posts: 152
Credit: 3,542,579
RAC: 5
United States
Message 139617 - Posted: 20 Jul 2005, 15:20:55 UTC - in response to Message 139600.  

I wonder why they do not at least upgrade to 4.19 :-/


Most BOINC/SETI participants come from the Ronco philosophy of computer management. ie "Set it and forget it!" [audience claps]

By the way, those things don't work, they don't have enough mass to maintain an even cooking temperature. ;) Whenever I'm cooking in the oven I always leave a couple of pizza stones on the bottom to help maintain an even temperature inside the oven.

7,049 S@H Classic Credits
ID: 139617 · Report as offensive
Bart Barenbrug

Send message
Joined: 7 Jul 04
Posts: 52
Credit: 337,401
RAC: 0
Netherlands
Message 139655 - Posted: 20 Jul 2005, 16:55:13 UTC
Last modified: 20 Jul 2005, 16:56:15 UTC

I agree with Ananas. here's another example of a WU with three succesful results (so it could've been validated ok), but the WU got trashed since too many download errors came first (only two out of the 6 are by pre-4.19 core clients by the way, so not sending work to pre-4.19 clients wouldn't even have helped in this case).

I don't see why a result could not transition from "Ready to send" to "In progress" (to use the status page terminology) only *after* it's been succesfully downloaded by a client. If an error occurs, fine, just keep it as "ready to send" and hand it out to the next client who comes asking for more work. No need to keep track of the failed attempt in the database: the client will come again for more (just like a failed upload of a processed result will simply be retried later), so there is no strain (even less than in the current way of working) on the database. Am I missing something?
ID: 139655 · Report as offensive
Profile Paul D. Buck
Volunteer tester

Send message
Joined: 19 Jul 00
Posts: 3898
Credit: 1,158,042
RAC: 0
United States
Message 139677 - Posted: 20 Jul 2005, 17:30:49 UTC - in response to Message 139655.  

Am I missing something?

No, and I made the suggestion for a change in the policy. But, unless a project gets adamant about this as a change, it is not likely to change in the near term.
ID: 139677 · Report as offensive
SeaEagle

Send message
Joined: 14 Jun 99
Posts: 12
Credit: 3,291,985
RAC: 2
United States
Message 139697 - Posted: 20 Jul 2005, 17:49:01 UTC - in response to Message 139617.  

That's the best cooking tip i've heard in a long time, as soon as i'm done typing i'm putting mine in the oven, thanks.

Gregg
ID: 139697 · Report as offensive
ampoliros
Volunteer tester
Avatar

Send message
Joined: 24 Sep 99
Posts: 152
Credit: 3,542,579
RAC: 5
United States
Message 139730 - Posted: 20 Jul 2005, 18:48:19 UTC - in response to Message 139697.  

That's the best cooking tip i've heard in a long time, as soon as i'm done typing i'm putting mine in the oven, thanks.


You're too kind, I just said that it helps with maintaining an even temp (no wild fluctuations when the coils turn on and off or you open the door), I never said anything about how my cooking tastes.

You want to learn how to cook and not just how to say "BAM!" check out Alton Brown.

7,049 S@H Classic Credits
ID: 139730 · Report as offensive
Ingleside
Volunteer developer

Send message
Joined: 4 Feb 03
Posts: 1546
Credit: 15,832,022
RAC: 13
Norway
Message 139780 - Posted: 20 Jul 2005, 20:06:05 UTC - in response to Message 139600.  

I think there's a better solution than forcing people to install different versions.

The handling of download errors is just wrong. A download error is _not_ an error result, it isn't a "result" at all. So it should not count in "Too many error results".

After each download error, the next delivery of the same work unit should be delayed but no further action should be taken.

The download errors should not even be stored in the database for long, they are usually caused by technical server problems so they are totally irrelevant for the project. Make note about the error in some log file and forget it would be more appropriate.


If there are way more DL errors with 4.13 than with 4.45 that might also mean that 4.45 isn't as popular as 4.13, if 400 try with 4.13 and 100 try with 4.45, it is only normal that there are more errors with 4.13 in such a phase with server problems. It would be interesting to count maybe over a month, how many results come in from which core version.


Download-error can also be due to wu getting corrupted on disk, wrongfully deleted from disk, or wrongly signed on either wu or application. If you don't count any error-messages reported due to downloading, this means a potentially bad wu can be re-issued infinite many times and the whole point of counting errors is to guard against this.


As for more download-errors with v4.13, it's a known bug that if for any reason a download doesn't happen, it's never re-tried at all.
ID: 139780 · Report as offensive
Mibe, ZX-81 16kb
Volunteer tester

Send message
Joined: 30 Jun 99
Posts: 42
Credit: 2,622,033
RAC: 0
Sweden
Message 139783 - Posted: 20 Jul 2005, 20:09:09 UTC - in response to Message 139677.  


No, and I made the suggestion for a change in the policy. But, unless a project gets adamant about this as a change, it is not likely to change in the near term.


So what can be done to get the project aware of the importance of this problem?

It is an issue in a number of threads both here and in the problem forum allready.

And since the solution will lessen the burden on the servers, a bunch of other network problems is taken care of in the same fix.
ID: 139783 · Report as offensive
krgm
Volunteer tester

Send message
Joined: 2 Jun 05
Posts: 30
Credit: 72,152
RAC: 0
Canada
Message 139909 - Posted: 21 Jul 2005, 1:28:22 UTC

Now, I was just looking at my results. 2 of them have 0 credits due to this problem. All the download errors were reported before my computer had started work on the WU's. Can SETI not send a stop work order (or something like that) to BOINC to not persue further work on that WU?
ID: 139909 · Report as offensive
Don Erway
Volunteer tester

Send message
Joined: 18 May 99
Posts: 305
Credit: 471,946
RAC: 0
United States
Message 139916 - Posted: 21 Jul 2005, 2:00:17 UTC

All they need do, is treat "download" errors differently.

Just keep trying more hosts, until you get enough hosts that *successfully* download it. If you get 4 straight failures, maybe it is a network outage, or maybe it is a bad WU.

This would ensure that enough hosts always actually start on each WU, whereas now, the system allows hosts to grind away on WUs that it should know, apriori, will be useless.

Don

ID: 139916 · Report as offensive
Previous · 1 · 2 · 3 · 4 · Next

Message boards : Number crunching : Damn You People Using Boinc 4.13


 
©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.