Message boards :
Number crunching :
Has Seti changed modus operandi?
Message board moderation
Author | Message |
---|---|
kittyman Send message Joined: 9 Jul 00 Posts: 51469 Credit: 1,018,363,574 RAC: 1,004 |
Please help to clear up my confusion at the moment. I have been reading things in various threads since the mass exodus of WUs from users hosts yesterday. Has the basic way Seti grants credit changed? Is it now first 2 outta 3 with the 3rd not getting credit? Is the server gonna auto-delete the WU from the 3rd host as soon as the first 2 are in? Was this a one-time fix for the ghost WUs or an ongoing change? Do we have an official post from Matt or Eric as to what is going on? Forgive me if I have missed something somewhere, but I have not found answers to these questons yet. And the kitties are puzzled......... "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Keith T. Send message Joined: 23 Aug 99 Posts: 962 Credit: 537,293 RAC: 9 |
Please help to clear up my confusion at the moment. Only Eric or Matt or one of the other project staff can answer this authoritatively, but from what I have seen and read, it appears that only results not yet started are being cancelled by the project. You should not see any results that are partly crunched or completed being cancelled. Keith |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
I'm not sure at this point. My read of Eric's post in Tech News was this is more of a test of the features and how they work in the SAH environment rather than a permanent policy change at this point. This makes sense to me since there is a fairly large number of known ghosts and other anomalous results currently in the pipeline, so you get a lot of data back for a lot of different scenarios. OTOH, it seems clear there are a few problems with the way it works (which I'm assuming is the default for the functions, more or less) in the SAH environment as it stands. The one issue which is really troublesome is the hosts which have had a result started and/or completed and then aborted or rejected by the project. My understanding of the functionality is that should only happen if the project had actually canceled the WU. IOW, it shouldn't have any effect on work which is already running or completed. Alinator |
kittyman Send message Joined: 9 Jul 00 Posts: 51469 Credit: 1,018,363,574 RAC: 1,004 |
Please help to clear up my confusion at the moment. So is the 3rd result reported still going to get credit? Is it any WU not started that are being scuttled if they are the 3rd, or just WUs determined to be ghosties? "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Stoo Send message Joined: 26 May 99 Posts: 52 Credit: 455,941 RAC: 0 |
I believe so, I've seen a few in my history where there are three results with credit for a WU, others with only two and a cancelled type state for the third. |
ohiomike Send message Joined: 14 Mar 04 Posts: 357 Credit: 650,069 RAC: 0 |
My only major concern is the effect on the host's WU Quota. ie: is our quota (wu cache size) going to be reduced as it normally is for bad results? Boinc Button Abuser In Training >My Shrubbers< |
Keith T. Send message Joined: 23 Aug 99 Posts: 962 Credit: 537,293 RAC: 9 |
Please help to clear up my confusion at the moment. I don't think the server can tell if a result is a ghost or not, unless the new feature that Eric has enabled does this. I have received credit for one 3rd result today http://setiathome.berkeley.edu/workunit.php?wuid=133301172 I have another 3rd result currenly crunching http://setiathome.berkeley.edu/workunit.php?wuid=133626807 which will report in the next few hours (it's currenntly prempted @44.3% by a Beta WU). I olny have 9 results in progress http://setiathome.berkeley.edu/results.php?hostid=3001110 but I have had 4 aborted by project, 3 yesterday and 1 today. My time zone is UTC+1. Sir Arthur C Clarke 1917-2008 |
Philadelphia Send message Joined: 12 Feb 07 Posts: 1590 Credit: 399,688 RAC: 0 |
Please help to clear up my confusion at the moment. I've been the 3rd reporter on WU's and I've received the credit. They were ones that only required 2 reports also. Here are a one example from both of my computers where I was the third reporter of the WU: http://setiathome.berkeley.edu/workunit.php?wuid=131542582 http://setiathome.berkeley.edu/workunit.php?wuid=131669239 By the way, while I was looking for them and going through my completed WU's, I noticed an extremely high number of client errors on other computers. I don't think they are because of them being the third (based on me getting mine), I think there is some other issue going on. |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
Sure it can, the host reports what it has onboard and the scheduler already knows what should be onboard. 4.45 (IIRC) and later clients have the report back functionality. The project side just doesn't do anything with the info unless the feature is enabled. FWIW, I have a cache setting of 1 day for my Intels and I haven't had any of my results aborted or my trailers not accepted by the project as of yet. Alinator |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
... Same here, cache setting of 1.2 days, crunching away on resend shorties. "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
Henk Haneveld Send message Joined: 16 May 99 Posts: 154 Credit: 1,577,293 RAC: 1 |
The aborts after 2 results have been returned happen on BOINC version 5.8.17 and higher as posted in a other message. |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
Agreed, but ghost resends were what we we're talking about. Although I see what you are saying about mine and since I follow the dev forums as well is one reason I haven't gone higher than 5.8.15 yet. Alinator <edit> I suppose one could say you early adopters of 5.8.17 and higher got 'volunteered' to be the guinea pigs for this experiment. ;-) Alinator |
zombie67 [MM] Send message Joined: 22 Apr 04 Posts: 758 Credit: 27,771,894 RAC: 0 |
By the way, while I was looking for them and going through my completed WU's, I noticed an extremely high number of client errors on other computers. I don't think they are because of them being the third (based on me getting mine), I think there is some other issue going on. Are you sure? When the server aborts results on my machine, they show up as "client error" when returned to the server. Dublin, California Team: SETI.USA |
Philadelphia Send message Joined: 12 Feb 07 Posts: 1590 Credit: 399,688 RAC: 0 |
By the way, while I was looking for them and going through my completed WU's, I noticed an extremely high number of client errors on other computers. I don't think they are because of them being the third (based on me getting mine), I think there is some other issue going on. I agree that the "client error" is an abort of some kind but I don't believe it is because it is the "third" WU being reported of three (now that only two are required). All of my WU's that were the "third" of three reported received cobblestones. I think those "client errors" are related to something else. |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
For Auto-aborts, based on what I've seen so far: Server State is 'Over', Outcome will be 'Client Error', Client State will show 'Aborted'. Anything else is some other problem, although I suppose a user abort might show the same thing depending on CC version. Also as has been mentioned, only 5.8.17 and higher will honor the auto abort command. Alinator |
Keith T. Send message Joined: 23 Aug 99 Posts: 962 Credit: 537,293 RAC: 9 |
By the way, while I was looking for them and going through my completed WU's, I noticed an extremely high number of client errors on other computers. I don't think they are because of them being the third (based on me getting mine), I think there is some other issue going on. BOINC FAQ Service explains the error codes. ERR_ABORTED_BY_PROJECT -221 The project you are attached to aborted this result. ERR_ABORTED_VIA_GUI -197 You have aborted the work unit through Boinc Manager The Exit status shows up on the Result page but not on the Results for computer page. Sir Arthur C Clarke 1917-2008 |
zombie67 [MM] Send message Joined: 22 Apr 04 Posts: 758 Credit: 27,771,894 RAC: 0 |
I agree that the "client error" is an abort of some kind but I don't believe it is because it is the "third" WU being reported of three (now that only two are required). All of my WU's that were the "third" of three reported received cobblestones. I think those "client errors" are related to something else. I watched BOINC abort a WU, then upload & report it. When I looked at the result from my account page, it shows as client error. Now, why was the result aborted by the server? No idea. Dublin, California Team: SETI.USA |
Philadelphia Send message Joined: 12 Feb 07 Posts: 1590 Credit: 399,688 RAC: 0 |
I agree that the "client error" is an abort of some kind but I don't believe it is because it is the "third" WU being reported of three (now that only two are required). All of my WU's that were the "third" of three reported received cobblestones. I think those "client errors" are related to something else. I honestly don't know why that would be but as I mentioned in my earlier post #584589, I showed two WU's, one from each of my computers, where each one was the third of three to report and in both cases only two were required for verification, in both cases I recieved cobblestones and neither were reported as "client error". |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
@ Zombie: Check your exit code from the Result detail page. All the ones I saw were -221 Auto-aborts. @ Philly: You're running 5.8.15 on your hosts. It doesn't honor the Auto-abort command, so it will run any result it DL's regardless of whether it's currently a trailer or not. Alinator |
zombie67 [MM] Send message Joined: 22 Apr 04 Posts: 758 Credit: 27,771,894 RAC: 0 |
@ Zombie: Yep, mine are all -221. My point was, I have no idea why the server decided to aboard them. Dublin, California Team: SETI.USA |
©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.