Message boards :
Number crunching :
Lost/Missing Workunits
Message board moderation
Author | Message |
---|---|
Cruncher-American Send message Joined: 25 Mar 02 Posts: 1513 Credit: 370,893,186 RAC: 340 |
When I query my account from BOINC Mgr (6.6.20) Projects tab for SETI@home/Your Results, I see over 100 wu as "In Progress". My machines are both 8-core machines, and have about 30 or so wu in queue. And I process more than that every day. So there are about 30 or so listed as "In Progress" that I know nothing about, and am confused about why they are "In Progress". (They are the ones more than a day old). How can I find out why they are there? And what can I do about them? They are obviously holding up credit for the other users that I am paired with on them. Some of them may be due to a power outage I had a few days ago... What info do you need from me to track them down? Thanks, Jon Ravin |
[B^S] madmac Send message Joined: 9 Feb 04 Posts: 1175 Credit: 4,754,897 RAC: 0 |
|
Cruncher-American Send message Joined: 25 Mar 02 Posts: 1513 Credit: 370,893,186 RAC: 340 |
Actually, it turned out to be my bad (with one exception) - or, as Roseanne Rosannadanna used to say, "Nevermind". On May 31 and June 1 I switched my machines to the SSE3 64bit Optimized MB App (which increased my throughput by about 75%, by the way). At the time, I lost all of the WU I had in my queues. They just disappeared... (I hadn't realized that would happen). The WU from those 2 dates in my "In Progress" list are (I assume) those missing WU. There is also one dated May 21; maybe that's a "ghost". Is there any way of returning the lost WU to my queues? Or at least marking them as available in the database, since they are not going to get finished until they are distributed to other users late this month, if left untouched? |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14653 Credit: 200,643,578 RAC: 874 |
The only way that I know of is to detach the host from SETI, then re-attach. You'll probably want to run your current cache down to nothing first - no point in causing a resend for work which you do have available to crunch. I'm planning to do that this afternoon to release the ~150 tasks which the faulty BOINC v6.6.34 gobbled up - I'll let you know how I get on. |
Cruncher-American Send message Joined: 25 Mar 02 Posts: 1513 Credit: 370,893,186 RAC: 340 |
Thanks for the suggestion. Please let me know if it works! |
Ray Send message Joined: 16 Jun 99 Posts: 30 Credit: 1,323,477 RAC: 0 |
I have just checked one of your pc's (4912909) and you have a total of 531 results (Details), 37 (In progress), 313 (Pending), 153 (Valid) and 7 (Error). As of 11:21 a.m. Eastern time. The only work units you see under BOINC will be the work units that are either waiting to be processed and ones that have been processed but not reported. The ones "In Process" are ones that you're pc has yet to do,"Pending" are the ones you have completed but your wing man has yet to do. "Valid" are the ones that have been verified by the SETI server. The "Your Results" button in the BOINC Mgr (6.6.20) is showing all the results from all 3 of your pc's that are attached to SETI. Hope this gives you a bit of clarification into things. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14653 Credit: 200,643,578 RAC: 874 |
Thanks for the suggestion. Please let me know if it works! Yes, it worked - you can see the results on host 3755243. Notes: 1) If you use any optimised applications, BACK UP YOUR PROJECT FOLDER. It gets deleted when you detach. Best order is complete last task --> report completed tasks (using project update button) --> back up project directory --> detach --> restore project directory --> re-attach. 2) It's b****y difficult to get three CPU cores and a CUDA card to run out of work at the same time! 3) And it's best not to try to do it on a weekend when the servers are playing up particularly badly.... 4) As you can see, I got my old Host ID back (no special action on my part). But you'll have to look for the 'client detached' results yourself (tasks issued 4 June and before), because none of the filters will find them. 5) Work estimates, cache size etc. will be screwy for a little while. The first CUDA task after the re-attach was a shorty: estimated over an hour, completed in 7m:30s as usual. |
Cruncher-American Send message Joined: 25 Mar 02 Posts: 1513 Credit: 370,893,186 RAC: 340 |
I have just checked one of your pc's (4912909) and you have a total of 531 results (Details), 37 (In progress), 313 (Pending), 153 (Valid) and 7 (Error). As of 11:21 a.m. Eastern time. Ray - thanks; I had already figured most of this out. There are some "In Progress" ones that are from 1 June and 31 May that are the ones I am having a problem with, and were the ones I originally asked about that were blown away by changing to the optimized app = apparently BOINC or SETI discards the in queue or not completed WUs when changing to the Opt. apps. Jon |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 |
... It's a matter of matching the app version in the upgrade to the version chosen to do the work when it was downloaded. And an "app version" for BOINC 6.6.x actually includes 4 different criteria; the name, number, platform, and plan_class. Your transition was from stock 6.03 (which is a 32 bit application) to the 64 bit optimized version, there has been some uncertainty about how to handle the effective platform change. For most users upgrading from stock to an optimized app there shouldn't be a loss of work, but not all transitions have been exhaustively tested. Each test would involve transitioning back to stock first and that's at least a nuisance. Joe |
Gads Send message Joined: 3 Dec 99 Posts: 11 Credit: 1,356,605 RAC: 0 |
I was working through my first AstroPulse assignment and about 4 hours from completion when I upgraded my hard drives and was unable to restore the image I'd created to the new drive. As a result I lost the AstroPulse work and several Multi-Beam tasks waiting to run and will not be able to return them. Will a failure to return the AstroPulse work unit by the deadline disqualify me from running them in the future because of a perceived inability to process the data in a timely manner? It appeared that one core in my X4 processor could crunch a work unit in about 48 CPU hours so once the pipeline was filled I could probably return about one or so complete units each day. |
1mp0£173 Send message Joined: 3 Apr 99 Posts: 8423 Credit: 356,897 RAC: 0 |
Will a failure to return the AstroPulse work unit by the deadline disqualify me from running them in the future because of a perceived inability to process the data in a timely manner? No. |
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
I was working through my first AstroPulse assignment and about 4 hours from completion when I upgraded my hard drives and was unable to restore the image I'd created to the new drive. As a result I lost the AstroPulse work and several Multi-Beam tasks waiting to run and will not be able to return them. Will a failure to return the AstroPulse work unit by the deadline disqualify me from running them in the future because of a perceived inability to process the data in a timely manner? It appeared that one core in my X4 processor could crunch a work unit in about 48 CPU hours so once the pipeline was filled I could probably return about one or so complete units each day.You have the phenom 9500. If you use the optimized app for SSE3 astropulse you'll be finishing AP WU's in about 24-27 hours In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
©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.