Message boards :
Number crunching :
Abandoned Tasks
Message board moderation
Previous · 1 · 2
Author | Message |
---|---|
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
You could try the "report the same task twice" trick, when requesting new work Talk to me like a newbie please, how am I supposed to do that? :) |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
You could try the "report the same task twice" trick, when requesting new work Umm. Claggy's the expert, and he's offline at the moment. I'll see if I can remember. I think there are two ways: 1) (probably best to set 'NNT' for this one): report completed task(s) as normal, but pull the network cable before you get the 'scheduler reply' acknowledging them. Wait five minutes, unset NNT, and plug the cable back in. As Jeff says, it's probably best if you've been on NNT for a while to make space in your cache. At your network speeds, it may be difficult to pull the plug at exactly the right moment, so 2) Don't set NNT, so tasks don't report immediately after upload. Maybe reduce cache instead, so they hang around for a while. Find a task which has uploaded and is ready to report. Look in client_state.xml (you can do this while BOINC is running) and find both the <workunit>...</workunit>, and <result>...</result>, sections for that workunit. The <result> section will have the full stderr report, which helps to ensure you've got the right one. Copy both sections (make sure the workunit names match) to a temporary text file for safe keeping. Report all completed tasks. Stop BOINC, open client_state in edit mode, and put back the two sections you just copied (make sure you put them in the SETI project section). Save the file, and restart BOINC. Once the five minutes have passed (bug - I think the client is resetting the server-requested backoff again - we fixed that), bump your cache size a notch or two to trigger a work request. You might like to wait for a few minutes to allow Jeff or anybody else to correct any details I've forgotten. |
Jeff Buck Send message Joined: 11 Feb 00 Posts: 1441 Credit: 148,764,870 RAC: 0 |
You could try the "report the same task twice" trick, when requesting new work Here's the technique that worked for me, though it's not definitive and may not be the most efficient approach. 1. Wait until you have a single task that has finishing uploading and is ready to report. 2. Suspend processing, suspend network activity, and set NNT. 3. Make a backup copy of BOINC directory. 4. Resume network activity, then "update" the project to report the waiting task (although, with NNT set, that may happen automatically). 5. Exit BOINC completely. 6. Restore the BOINC directory backup. 7. Restart BOINC. 8. Increase your work buffer to accommodate your "lost" tasks. 9. "Allow new tasks", then resume network activity. 10. "Update" the project to once again report the same waiting task. This "should" result in one or more "lost" tasks being resent. 11. Resume normal processing. I hope I haven't left anything out. :^) If, as Richard mentioned, there's a 20 task resend limit, you might have to try this a couple times. Good luck! EDIT: Changed the order in Step 9 to put "Allow new tasks" before resuming network activity. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
10. "Update" the project to once again report the same waiting task. This "should" result in one or more "lost" tasks being resent. Didn't even get the chance to hit Update, it auto-reported as soon as I took it off Suspend Network. Now, I don't see BOINC stating the server resent anything, but I do notice that I have all kinds of tasks dated for 12 February back, where before all work was for 16 March. So not sure if this worked, or not. Can't do the trick with the network cable, as it's behind my computer on the floor under my desk and by the time I'm there you can call an ambulance... One tip for the next following this, make sure your BOINC isn't set to use <report_results_immediately> :) |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
Well.....speak of the devil and he arrives.... I actually DO 3 see v8 error tasks today. But, they are the comms glitch 'resent with little time to crunch them' kind, so still nothing app related. Still all good in kittyland. Meow! "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Rasputin42 Send message Joined: 25 Jul 08 Posts: 412 Credit: 5,834,661 RAC: 0 |
3. Make a backup copy of BOINC directory. I believe, just copy the "Client_state.xml" is sufficient. You do not need all of the Boinc dir. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
I kinda like my method; 1) reduce cache setting to make room, say 0.5 days 2) when there is room, wait until a task is within a minute or 2 from completing then hit the manual Update button to give you 5 minutes 3) set cache to say, 1 day 4) as soon as the task finishes and uploads, copy the client_state_prev.xml & client_state.xml to a different directory 5) hit the Update button to report the task and Quit BOINC as soon as it finishes reporting 6) replace current client_state_prev.xml & client_state.xml with copied older versions 7) restart BOINC and the task will be reported again triggering the event 8) reset cache to whatever is appropriate Oh, and you Must have your preferences set to receive the ghost task, using "...accept work from other applications?" Won't work. The specific application must be selected. |
Jeff Buck Send message Joined: 11 Feb 00 Posts: 1441 Credit: 148,764,870 RAC: 0 |
10. "Update" the project to once again report the same waiting task. This "should" result in one or more "lost" tasks being resent. For me, the Event Log actually showed the lost tasks as being resent, but maybe in some circumstances it doesn't. Can't do the trick with the network cable, as it's behind my computer on the floor under my desk and by the time I'm there you can call an ambulance... One tip for the next following this, make sure your BOINC isn't set to use <report_results_immediately> :) Ah, never thought about that <report_results_immediately> setting. That would probably have the same effect as setting NNT before suspending the network activity. |
Jeff Buck Send message Joined: 11 Feb 00 Posts: 1441 Credit: 148,764,870 RAC: 0 |
3. Make a backup copy of BOINC directory. You may be right. I didn't try to figure out which scalpel to use, I just pulled out the hacksaw. It seemed simpler at the time. ;^) |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Can't do the trick with the network cable, as it's behind my computer on the floor under my desk and by the time I'm there you can call an ambulance... Tip: most cables have two ends. Mine are plugged into the back of the computers under the bench, too - but the other ends are plugged into a switch on top of the bench, between the monitors. One tip for the next following this, make sure your BOINC isn't set to use <report_results_immediately> :) What's that? I thought you said you were a newbie ;) |
Jeff Buck Send message Joined: 11 Feb 00 Posts: 1441 Credit: 148,764,870 RAC: 0 |
10. "Update" the project to once again report the same waiting task. This "should" result in one or more "lost" tasks being resent. Now that I look back at your earlier post, I see that you said "Funny side thing is that the tasks are actually also still in the ..\projects\setiathome.berkeley.edu\ folder.". That would probably mean that the tasks didn't actually have to be physically resent. In that case, I would think you'd still see lines in your Event Log such as "File xxxxxx exists already, skipping download". They'd still get added back into BOINC's task list, though. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
Can't do the trick with the network cable, as it's behind my computer on the floor under my desk and by the time I'm there you can call an ambulance... That would require me to run out the door, through the whole living room, trying not to trip over one of my cats who just then decides it's bathing time there. Then quickly figure out which of the four cables going into my router does not disconnect my new NAS, Holly's computer or the TV-server, but just mine without wrecking just about everything on the little cupboard I built for the modem, router and telephone. Yeah, let's not. I can do without the added stress. :) It's only two hands full of tasks. If they aren't done by me, they'll be done by someone eventually. I thought you said you were a newbie ;) Aren't we all a newbie at one point? Mine was for the "double reporting", not for much of the inner workings of BOINC. |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
Seeing how I have several tasks with a save-date of 7, 8, 9 and 12 October 2011 (!), and more of them from 2014 and 2015 in my ..\projects\setiathome.berkeley.edu\ folder, I wonder how much more crap I have in here. Oh look, 8 APs, while I have only got one. I think it's time to set NNT, run down the cache, report all, exit BOINC and do a manual clean up. Erm that is, this project hasn't changed to Locality Scheduling, has it? |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
Had one old task in my folder despite running down to zero, deleted that. "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. |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
Erm that is, this project hasn't changed to Locality Scheduling, has it? Not as far as I know, no. "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. |
BilBg Send message Joined: 27 May 07 Posts: 3720 Credit: 9,385,827 RAC: 0 |
Seeing how I have several tasks with a save-date of 7, 8, 9 and 12 October 2011 (!), and more of them from 2014 and 2015 in my ..\projects\setiathome.berkeley.edu\ folder, I wonder how much more crap I have in here. Oh look, 8 APs, while I have only got one. BOINC Rescheduler can search ([Test for lost files] button) and delete ([Delete lost files] button) but only SETI@home (MB) tasks, not for AstroPulse tasks It checks for files with 3 or more dots in the filenames and looks if they are listed in client_state.xml - I just tested with a fake file: 27 January 2016 - 10:12:28 Checking for lost files 27 January 2016 - 10:12:28 File not found: gggame.lo.g.txt ,this file should be deleted. 27 January 2016 - 10:12:28 Found: 1 files that are lost and need to be deleted 27 January 2016 - 10:12:41 Checking for lost files 27 January 2016 - 10:12:41 File not found: H:\BOINC-Data\projects\setiathome.berkeley.edu\gggame.lo.g.txt ,deleting. 27 January 2016 - 10:12:41 Found: 1 files that are lost and need to be deleted 27 January 2016 - 10:12:41 Deleted: 1 files  - ALF - "Find out what you don't do well ..... then don't do it!" :)  |
©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.