One CPU Intel Pentium 4 2.40GHz with 1579 tasks in progess!
Message boards :
Number crunching :
One CPU Intel Pentium 4 2.40GHz with 1579 tasks in progess!
Message board moderation
| Author | Message |
|---|---|
 red-ray  Send message Joined: 24 Jun 99 Posts: 308 Credit: 9,029,848 RAC: 0 
|
I did not think there were any AP WUs, so when I just got one I looked up it's history and found http://setiathome.berkeley.edu/show_host_detail.php?hostid=698369 which is a 1 CPU P4 System, with a RAC of 4.87 and 1579 tasks in process! How come it can get that many when there is supposed to be a limit of 50 per CPU? Systems such as this need to be blacklisted. Is this possible? ID: 1203465 ·  |
Gatekeeper  Send message Joined: 14 Jul 04 Posts: 887 Credit: 176,479,616 RAC: 0 
|
Might have something to do with the BOINC version (4.25) on this rig. That's all I can see that is "unusual".  
ID: 1203467 · |
LadyL  Send message Joined: 14 Sep 11 Posts: 1679 Credit: 5,230,097 RAC: 0 |
Old BOINC versions don't report the tasks they have on board in their scheduler request. Since we know that ghosts don't count towards the quota, I'd wager it counts what the host reports as having on board - and if that is empty... I'm not the Pope. I don't speak Ex Cathedra! ID: 1203482 · |
|
red-ray Send message Joined: 24 Jun 99 Posts: 308 Credit: 9,029,848 RAC: 0
|
Old BOINC versions don't report the tasks they have on board in their scheduler request. Since we know that ghosts don't count towards the quota, I'd wager it counts what the host reports as having on board - and if that is empty... How old does the BOINC have to be? It's about time reqests from at pre version 5 got rejected. I wonder how many other such systems there are. ID: 1203484 · |
Richard Haselgrove  Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874
 |
I can't find a timeline for clients as old as that - but we were already up to BOINC v5 when I switched from classic in November 2005. I bet that server hasn't been touched since BOINC was installed in April 2005 - maybe the system builder installed BOINC and moved away. It's not showing any crunched WUs at the moment, but it must have done some reasonably recently, to have an appreciable RAC. Somebody is stumping up for a lot of AP download bandwidth. Edit - he's 6th. out of 206 for his sign-up date, with just that one computer, ROFL. BOINCstats ID: 1203486 · |
Khangollo  Send message Joined: 1 Aug 00 Posts: 245 Credit: 36,410,524 RAC: 0 
|
These workunit "black holes" are not so uncommon. Forgotten computers that stockpile workunits endlessly (past limits) and never return them. I wish there was some kind of mechanism in place, to temporarily ban computers that continuously produce timeout errors. 
ID: 1203496 · |
|
red-ray Send message Joined: 24 Jun 99 Posts: 308 Credit: 9,029,848 RAC: 0
|
Old BOINC versions don't report the tasks they have on board in their scheduler request. Since we know that ghosts don't count towards the quota, I'd wager it counts what the host reports as having on board - and if that is empty... I just got the following and it make me wonder, does BOINC keep resending all the ghosts that aren't really ghosts to hostid=698369 and wasting bandwidth? If so this should be addressed. I feel hostid=698369 should be blocked totally (http://www.urbandictionary.com/define.php?term=send%20the%20boys%20round seems about right to me!). 08/03/2012 00:36:06 | SETI@home | Scheduler request completed: got 8 new tasks ID: 1203530 · |
ivan  Send message Joined: 5 Mar 01 Posts: 783 Credit: 348,560,338 RAC: 223
|
Old BOINC versions don't report the tasks they have on board in their scheduler request. Since we know that ghosts don't count towards the quota, I'd wager it counts what the host reports as having on board - and if that is empty... Do you actually get the file resent? I had a number of those last night where the next messages were "already have that file" with the implication that it wasn't re-downloaded. OTOH, I've also had a number a while back that did appear to really be ghosts and were finally downloaded. AFAIUI, "resent" just means "try downloading this one again"; the local BOINC then has the option of saying, "Nah, already go' i'!" or "Thanks, I'll try again."   
ID: 1203531 · |
|
Gatekeeper Send message Joined: 14 Jul 04 Posts: 887 Credit: 176,479,616 RAC: 0
|
Old BOINC versions don't report the tasks they have on board in their scheduler request. Since we know that ghosts don't count towards the quota, I'd wager it counts what the host reports as having on board - and if that is empty... IMX,a "resend" such as the above follows a "scheduler timed out" error on a previous D/L attempt. The scheduler asssigned X WU's to your system, but the advice never reached you. The next time you contact the scheduler, it looks at your existing task list, says "oops, you don't have these", and resends them in groups of 20 until all the work you "should have" arrives. When this feature was turned off awhile back, it wasn't unusual to see 300-500 "ghost" WU's aging out on your stats page, and you had no idea that they were never received, unless you did alot of cross checking actual inventory vs. S@H stats.
ID: 1203540 · |
|
red-ray Send message Joined: 24 Jun 99 Posts: 308 Credit: 9,029,848 RAC: 0
|
Old BOINC versions don't report the tasks they have on board in their scheduler request. Since we know that ghosts don't count towards the quota, I'd wager it counts what the host reports as having on board - and if that is empty... Thank you. That makes sense. What can be done to block hostid=698369 and similar system though? BTW is [4] NVIDIA GeForce GTX 590 4 GPUs or 8? ID: 1203542 · |
Wiggo  Send message Joined: 24 Jan 00 Posts: 34744 Credit: 261,360,520 RAC: 489 
|
What can be done to block hostid=698369 and similar system though? Once we're all forced to go to version 7 that will stop all of those systems. ;) Cheers. ID: 1203553 · |
|
red-ray Send message Joined: 24 Jun 99 Posts: 308 Credit: 9,029,848 RAC: 0
|
What can be done to block hostid=698369 and similar system though? Good idea. I propose 01-Apr-2012. ID: 1203555 · |
|
Gatekeeper Send message Joined: 14 Jul 04 Posts: 887 Credit: 176,479,616 RAC: 0
|
BTW is [4] NVIDIA GeForce GTX 590 4 GPUs or 8? 2 cards; 4 GPU's; 8 S@H tasks simultaneously.
ID: 1203556 · |
|
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0
|
I can't find a timeline for clients as old as that - but we were already up to BOINC v5 when I switched from classic in November 2005. http://www.boinc-wiki.info/Version_History_of_The_BOINC_Client_Software For cross-reference, here's the first part of the checkin-note: David 31 May 2005
- Add 2 new items to scheduler RPC request:
- <other_results>
List of other results for this project
(besides those being reported, if any)
This can be used by the scheduler to resend
previously sent results that were lost somehow.
Includes only the name of the result.
- <in_progress_results>
List of all results for which computation is not done.
Includes estimated remaining CPU time, report deadline.
This can be used by the scheduler to decide whether
to send short-deadline results.
JoeID: 1203557 · |
Link  Send message Joined: 18 Sep 03 Posts: 834 Credit: 1,807,369 RAC: 0 
|
I wish there was some kind of mechanism in place, to temporarily ban computers that continuously produce timeout errors. Well, there is and it seems to actually work for this machine if you look at how many MB WUs it got in the last few days: 1/day. On AP he's still at 33/day, so there we need to wait for few more to time out, than the quota system will hopefully work there also. Seems to be quite a new problem with this machine, 2-3 weeks ago it could still get quite many MB WUs per day, so it must have been still working OK about 2-4 months ago. 
ID: 1203626 · |
shizaru  Send message Joined: 14 Jun 04 Posts: 1130 Credit: 1,967,904 RAC: 0 
|
Doesn't have the same problem as your guy, but here's my favorite wingman from my list of pendings:) http://setiathome.berkeley.edu/results.php?hostid=3378825 ID: 1204152 · |
|
LadyL Send message Joined: 14 Sep 11 Posts: 1679 Credit: 5,230,097 RAC: 0 |
Doesn't have the same problem as your guy, but here's my favorite wingman from my list of pendings:) What the heck is he running?! oh r390 mhm. I hate those guys who run beta and hide their computers. looks like r390 was the last one that made it out of alpha... If he gets a few good units in inbetween , the quota mechanism will allow him some more. I'm not the Pope. I don't speak Ex Cathedra! ID: 1204155 · |
skildude  Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 
|
lol waiting to be invalid. In his case its very true  In a rich man's house there is no place to spit but his face. Diogenes Of Sinope ID: 1204544 · |
|
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0
|
3378825 is certainly an interesting case. It's sad to see a top host ranked 1067 by RAC and 882 by total credit not doing as well as it should. Raistmer announced r390 at SETI Beta last November and probably here too. From the BoincStats graphs for the host, its rate of total credit growth slowed about that time. That could just be coincidence, and the pending list doesn't go back far enough to see what it was running before then. If I had a system with 3 GPUs crunching, I'd be monitoring it fairly closely, but BOINC doesn't give a user much indication there's something wrong when only ~10% of tasks are validating. In this case the ~90% invalid fail very quickly so the RAC isn't reduced hugely. I've often thought there should be some obvious quality indication, but BOINC doesn't even keep error rate statistics. Joe ID: 1204700 · |
Dimly Lit Lightbulb 😀  Send message Joined: 30 Aug 08 Posts: 15399 Credit: 7,423,413 RAC: 1
|
3378825 is certainly an interesting case. It's sad to see a top host ranked 1067 by RAC and 882 by total credit not doing as well as it should. Maybe that could be added to Boinc, it keeps a record of the last (100, 1000, 10'000?) tasks and has a little graph added to the statistics tab (or one of it's own) showing validated vs. error rates. So if you see a spike of errors you know something has gone bonkers, either project wise via a visit to the forums or locally if no one else has reported anything.  Member of the People Encouraging Niceness In Society club. 
ID: 1204733 · |
©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.