Message boards :
Number crunching :
Any ideas on clearing Orphaned "results" please?
Message board moderation
Author | Message |
---|---|
Keith Send message Joined: 19 May 99 Posts: 483 Credit: 938,268 RAC: 0 |
I have the following "Results" (absent from "Tasks" page) just sitting waiting for their respective lapsing deadlines:- Result ID Work unit ID Sent deadline Name Comp’r ID 532369014 125023255 10 May 2007 22:39:53 UTC 3 Jun 2007 16:36:21 UTC 30ap04aa.5867.14145.473594.3.173_4 2680207 532369013 125023249 10 May 2007 22:39:53 UTC 3 Jun 2007 16:36:21 UTC 30ap04aa.5867.14145.473594.3.167_5 2680207 532097081 128452734 1 May 2007 22:39:39 UTC 26 May 2007 19:06:54 UTC 14fe05aa.7905.9297.934640.3.231_3 2680207 532126675 128633047 1 May 2007 23:05:28 UTC 18 May 2007 10:10:10 UTC 30se04aa.13975.30066.104832.3.54_0 2670914 532126625 128633043 1 May 2007 23:05:28 UTC 18 May 2007 10:10:10 UTC 30se04aa.13975.30066.104832.3.50_2 2670914 532126585 128633028 1 May 2007 23:05:28 UTC 18 May 2007 10:10:10 UTC 30se04aa.13975.30066.104832.3.36_2 2670914 Is there any way of aborting them so that others may have them re-issued? Thinking back to the old Classic SETI days, I wondered if there might be a possibility of creating a text file, renaming it, and placing it into the SETI subfolder in the BOINC Data folder. Knowing full well it would not run, but might just be sufficient to be recognised for abortion to take place. Tried it, but with no success even after changing permissions to "boinc_master". Anyone got any other suggestions. Would be nice to come from SETI, but they're probably too busy clearing stale crisps off the table during the outage. Keith |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
Hmmmm.... That's a clever idea. Create a 'fake' result for the ghost so BOINC would abort it and cause the project to reissue it. Thinking about it you would have to build the appropriate section for the ghosts in the client_state file. Then when BOINC tried to run it, it would abort when it couldn't find the input files. If any of your hosts still have any work for SAH running, their client_state file would give you a 'template' for what you need add to fake the ghost. <edit> Thought just occurred to me the main BOINC site might have enough info to be able to 'manufacture' the result section for client_state, even if you don't have a template available. It probably would not be strictly 'correct', but might be good enough to force the abort, which is all you really want. ;-) HTH, Alinator |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 |
I have the following "Results" (absent from "Tasks" page) just sitting waiting for their respective lapsing deadlines:- By design, it's nearly impossible to convince the BOINC core client that work has been done when it hasn't. There is a feature in the BOINC system designed to handle such cases by resending the work to the host when the records don't match. S@H normally doesn't have that feature enabled, but I just sent an email to Eric suggesting he might consider enabling it while there isn't any new work available. I haven't researched how much impact on the BOINC database and Scheduler it would be, so have no idea whether it's a practical idea. Joe |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
Agreed, but I wasn't thinking about trying to fake BOINC into believing the result had been actually completed, just that it had the ghost in the bullpen waiting to run. And you're definitely correct the 'real' soultion is to just turn on auto reissue for SAH, if practical, to render ghosts a moot point. ;-) Alinator |
Keith Send message Joined: 19 May 99 Posts: 483 Credit: 938,268 RAC: 0 |
Alinator & Jo Try to pull all possible strings to get that "auto issue" actioned. After all, once the new Sun is up and running, these wretched blemishes on the SETI database will survive beyond this ideal moment for a clean restart. It's an opportunity not to be missed. I unfortunately have no "results" at all to process, thanks to my keeping a reasonably small cache that would not hold up other hosts. Keith |
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0 |
LOL.... Well I don't how much 'pull' I have at SSL, but I would bet that Joe has a lot more than me! ;-) One thing for sure though is I don't think we'll see auto-reissue turned on when Thumper is brought back online, at least initially. I'm sure the team will give it another round of consideration when they do the 'debriefing' after they clear the outage. <edit> Actually, it's probably better to refer to the feature as 'auto-resend' rather than 'auto-reissue', just to avoid confusion because they're really separate and distinct. Alinator |
Keith Send message Joined: 19 May 99 Posts: 483 Credit: 938,268 RAC: 0 |
SUCCESS. I'm so glad to have found a SOLUTION !!!!!!!!!!!!!!!!!!!! If you find you have Results listed as "In Progress" in the "Your Results" page, which are not listed in your "Tasks" page (and are known as "Orphaned" or "Ghost" results), you can get these resent almost instantly by detaching Setiathome for that host. It worked for me, and if everyone else does the same we should be able to clean up this database in a reasonably short length of time as there is then no 3 week delay until the lapse deadline before the resend to another eagerly awaiting host. You can then attach to SETI again using your same account number and password, with all the details still intact, also showing the previously troublesome results, now marked as "Detached" and harmlessly awaiting the deadline date. It would probably be, however, more sensible from a programmer's point of view if these were to be removed straight away. Here's the details of the result of my efforts on my affected host:- 532374067 126488989 11 May 2007 20:40:15 UTC 5 Jun 2007 18:01:39 UTC Over Client detached New 0.00 --- --- 532374066 126848398 11 May 2007 20:40:15 UTC 4 Jun 2007 20:32:05 UTC Over Client detached New 0.00 --- --- 532374065 126802462 11 May 2007 20:40:15 UTC 29 May 2007 8:15:16 UTC Over Client detached New 0.00 --- --- 532374064 126101410 11 May 2007 20:40:15 UTC 1 Jun 2007 8:50:56 UTC Over Client detached New 0.00 --- --- 532369014 125023255 10 May 2007 22:39:53 UTC 3 Jun 2007 16:36:21 UTC Over Client detached New 0.00 --- --- 532369013 125023249 10 May 2007 22:39:53 UTC 3 Jun 2007 16:36:21 UTC Over Client detached New 0.00 --- --- 532097081 128452734 1 May 2007 22:39:39 UTC 26 May 2007 19:06:54 UTC Over Client detached New 0.00 --- --- 532097080 128623565 1 May 2007 22:39:38 UTC 4 May 2007 21:09:16 UTC Over Success Done 30,974.51 62.40 62.40 531488049 128428705 30 Apr 2007 23:01:03 UTC 3 May 2007 19:46:00 UTC Over Success Done 27,651.62 53.02 53.02 531487988 128428693 30 Apr 2007 23:01:03 UTC 4 May 2007 1:37:51 UTC Over Success Done 27,469.15 53.02 53.02 And as you see from the first Workunit, the effect was immediate re-send (although strangely there is no indication above as to the actual time of the detachment, which logically might have replaced the lapse deadline:- 525479357 1754257 23 Apr 2007 4:46:13 UTC 5 May 2007 0:42:24 UTC Over Success Done 25,217.45 62.45 pending 525479358 3303575 23 Apr 2007 4:46:13 UTC 18 May 2007 2:07:37 UTC In Progress Unknown New --- --- --- 525479359 3218422 23 Apr 2007 4:46:13 UTC 18 May 2007 2:07:37 UTC Over Client detached New 0.00 --- --- 532374067 2680207 11 May 2007 20:40:15 UTC 5 Jun 2007 18:01:39 UTC Over Client detached New 0.00 --- --- 532376066 2294242 12 May 2007 8:51:21 UTC 6 Jun 2007 6:12:45 UTC In Progress Unknown New --- --- --- I really am feeling quite chuffed at finding what seems to be a solution that we, as Crunchers, may be able to help clean up this database. Keith |
Keith T. Send message Joined: 23 Aug 99 Posts: 962 Credit: 537,293 RAC: 9 |
SUCCESS. I'm so glad to have found a SOLUTION !!!!!!!!!!!!!!!!!!!! Warning, this procedure may have other side effects to your host stats. I tried this on my Athlon, (I backed up the "\\BOINC\\projects\\setiathome.berkeley.edu" folder first, as I have my KWSN downloads in that folder.) I then tried to re-attach to http://setiathome.berkeley.edu/ and appear to have lost my host stats for that PC. Sir Arthur C Clarke 1917-2008 |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
Good warning, I think from my recent upgrade of a machine, that when you either 'update', or report a workunit ( when we get some work of course) that because the boinc folder is the same it might 'take from where it left off' though. please keep posting findings :D "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. |
Keith Send message Joined: 19 May 99 Posts: 483 Credit: 938,268 RAC: 0 |
Keith T Thanks for warning to myself and to others who may be interested. I did follow the same procedure on another affected computer in my network with complete success, though. But I did not take a back-up beforehand in either case. Keith |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 |
... I don't think "resent" is a good description. What happens in this case is the BOINC core client tells the Scheduler the host has detached, and the Scheduler then changes any "In process" results to "Client detached". It's quite useful now once all work which the host knows about has been uploaded and reported, good catch! Everyone must understand it will release ALL "In process" results, not just ghost units. OTOH, if Eric does activate "resend lost work" you won't get those ghosts you've released. It worked for me, and if everyone else does the same... It is certainly a useful approach for the small fraction of S@H participants who read these forums. Getting the word out to "everyone else" seems impractical. Joe |
©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.