[phenixbb] Continue mr_rosetta after system crash?
Terwilliger, Thomas C
terwilliger at lanl.gov
Mon May 12 09:32:11 PDT 2014
You can restart from the end of the last completed step in mr_rosetta. Here is how:
At each stage, existing solutions are saved as a python "pkl" file and can be read back in to mr_rosetta with "mr_rosetta_solutions=xxx.pkl". These solutions can be displayed with "display_solutions=True". Existing solutions are stored as "mr_rosetta_solution" objects which keep track of the model and its history, the map_coefficients and labels, etc. These can be read in to mr_rosetta with the keyword "mr_rosetta_solutions=results.pkl" and used as inputs for subsequent runs, starting at any step that can use those solutions.
NOTE: You can re-start mr_rosetta only at the beginning of major stages (like "place_model", "rosetta_rebuild" etc)...but not in between.
Normally at the end of a major stage a .pkl file is written out with text like "type this to see all the results". You can almost always give your original command, the command "start_point=xxx" and "mr_rosetta_solutions=my_pickle_file.pkl" and it should then continue on from there.
(see also: http://www.phenix-online.org/documentation/reference/mr_rosetta.html#viewing-solutions-and-restarting-with-saved-solutions)
Let me know if that doesn't do it!
All the best,
On May 12, 2014, at 7:42 AM, Oliver Clarke wrote:
> Hi all,
> I was running a mr_rosetta job on an 8-core machine, and after 3 days of apparently fruitful computation (784/1000 initial models generated), my system crashed for an apparently unrelated reason (there are no errors in the rosetta logs - I suspect a brief power outage or something of the sort).
> I am guessing the answer is no, but I just thought I’d check whether anyone knows of a way to continue such a job from where it left off without having to start from scratch?
> phenixbb mailing list
> phenixbb at phenix-online.org
More information about the phenixbb