Message boards : Number crunching : Old robetta jobs re-running?
Author | Message |
---|---|
Timo Send message Joined: 9 Jan 12 Posts: 185 Credit: 45,649,459 RAC: 0 |
I am just curious about why seemingly old robetta predictions that show as completed are still spawning fresh work units? Example 1) One of my computers has a work unit called: rb_10_17_51162_97022__t000__0_C1_fa_wt_0.25_SAVE_ALL_OUT_IGNORE_THE_REST_309298_1840, and if I search for task ID 51162 on the Robetta Queue page, I find that that simulation (which was submitted back on October 17th (thus I'm sure this is the same simulation as the work unit ID also says rb_10_17_51162) shows a status of the simulation being 'Completed' Robetta Queue Link here: Job 51162 - October 17th (Robetta Queue - Completed Structure Prediction) ... Is this a mistake? I see a couple of other jobs in my queue that look to be from old 'Completed' robetta runs. Just want to make sure cycles aren't being spun for nothing. Tim |
David E K Volunteer moderator Project administrator Project developer Project scientist Send message Joined: 1 Jul 05 Posts: 1018 Credit: 4,334,829 RAC: 0 |
This shouldn't happen often. Thanks for heads up, I'll look into this. |
Timo Send message Joined: 9 Jan 12 Posts: 185 Credit: 45,649,459 RAC: 0 |
It may not be a mistake, its possible that the job simply had spawned n number of jobs long ago, but hit a quota of needed models before expected yet the old jobs simply stayed in the mix. Just thinking out-loud. Thanks for looking into it! :) |
Timo Send message Joined: 9 Jan 12 Posts: 185 Credit: 45,649,459 RAC: 0 |
Hi David, just thought I'd lend a hand by deep diving a bunch of my work units and finding other examples. Some of these robetta tasks look to be really old, some of them look to be newer, what they all have in common is that a) they are all 'Completed' in the Robetta queue and b) I received these work-units this morning (ie. super recently): Please note that the majority of rb_~ / robetta jobs look to be valid/active simulations, these are 9 out of maybe 30-40 that I checked. Note also that my target run-time is 8 hours and my network usage settings for my work queue is set to hold between 0.6 and 0.9 days of work on all 5 of the machines I run Rosetta on (basically 24/7). Here are the 9 jobs I found that are running models for completed simulations: ---------------------------------------------------------- Robetta job 50648 - Completed Robetta Job Creation Date: 09/19/2014 Workunit pulled down today (Nov-18th-2015): https://boinc.bakerlab.org/rosetta/workunit.php?wuid=700431300 ---------------------------------------------------------- Robetta job 53133 - Completed Robetta Job Creation Date: 01/23/2015 Workunit pulled down today (Nov-18th-2015): https://boinc.bakerlab.org/rosetta/workunit.php?wuid=700431291 ---------------------------------------------------------- Robetta job 50045 - Completed Robetta Job Creation Date: 08/08/2014 Workunit pulled down today (Nov-18th-2015): https://boinc.bakerlab.org/rosetta/workunit.php?wuid=700423405 ---------------------------------------------------------- Robetta job 50762 - Completed Robetta Job Creation Date: 09/26/2014 Workunit pulled down today (Nov-18th-2015): https://boinc.bakerlab.org/rosetta/workunit.php?wuid=700411844 ---------------------------------------------------------- Robetta job 52751 - Completed Robetta Job Creation Date: 01/02/2015 Workunit pulled down today (Nov-18th-2015): https://boinc.bakerlab.org/rosetta/workunit.php?wuid=700393980 ---------------------------------------------------------- Robetta job 52075 - Completed Robetta Job Creation Date: 11/28/2014 Workunit pulled down today (Nov-18th-2015): https://boinc.bakerlab.org/rosetta/workunit.php?wuid=700389078 ---------------------------------------------------------- Robetta job 60135 - Completed Robetta Job Creation Date: 10/26/2015 Workunit pulled down today (Nov-18th-2015): https://boinc.bakerlab.org/rosetta/workunit.php?wuid=700351753 ---------------------------------------------------------- Robetta job 60555 - Completed Robetta Job Creation Date: 11/12/2015 Workunit pulled down today (Nov-18th-2015): https://boinc.bakerlab.org/rosetta/workunit.php?wuid=700379537 ---------------------------------------------------------- Robetta job 55174 - Completed Robetta Job Creation Date: 04/18/2015 Workunit pulled down today (Nov-18th-2015): https://boinc.bakerlab.org/rosetta/workunit.php?wuid=700360925 ---------------------------------------------------------- |
David E K Volunteer moderator Project administrator Project developer Project scientist Send message Joined: 1 Jul 05 Posts: 1018 Credit: 4,334,829 RAC: 0 |
Thanks Timo, I've added logic to cancel the batch when the Robetta job completes. This should prevent jobs from running that have been queued after enough models are made. |
Timo Send message Joined: 9 Jan 12 Posts: 185 Credit: 45,649,459 RAC: 0 |
Thanks Timo, Wow, quick turn around David! By the way, check your u.washington.edu email, I sent you a coffee (starbucks gift card) for all your hard work and putting up with us whining volunteers on a daily basis. Hope it's not stepping on any ethical rules about gifts or whatever. You deserve it. Cheers! Tim |
David E K Volunteer moderator Project administrator Project developer Project scientist Send message Joined: 1 Jul 05 Posts: 1018 Credit: 4,334,829 RAC: 0 |
Thanks so much Timo. I see your gift (I deleted it thinking it was spam but found it in my trash bin just now). There is one issue with canceling jobs - if the jobs have already been queued on the client they will still run if the client does not communicate with our servers during that period. I'll see how this goes. |
Message boards :
Number crunching :
Old robetta jobs re-running?
©2024 University of Washington
https://www.bakerlab.org