I am following the hotgym prediction demo instructions from Matt's tutorial video (https://www.youtube.com/watch?v=S-0thrzOHTc) on a clean develop build checked out March 7th.
I am running ./swarm.py and it quits with the following error (this is just an excerpt of the full log http://pastebin.com/w0v1Ym7c) Successfully submitted new HyperSearch job, jobID=1021 <jobID: 1021> 2 models finished [success: 0; eof: 0; stopped: 0; killed: 0; ERROR: 2; ORPHANED: 0; unknown: 0] ERROR MESSAGE: Exception occurred while running model 2481: IOError(2, 'No such file or directory') (<type 'exceptions.IOError'>) Traceback (most recent call last): File "/home/vagrant/nupic/nupic/swarming/utils.py", line 430, in runModelGivenBaseAndParams (completionReason, completionMsg) = runner.run() File "/home/vagrant/nupic/nupic/swarming/ModelRunner.py", line 238, in run maxTimeout=readTimeout) File "/home/vagrant/nupic/nupic/data/stream_reader.py", line 209, in __init__ self._openStream(dataUrl, isBlocking, maxTimeout, bookmark, firstRecordIdx) File "/home/vagrant/nupic/nupic/data/stream_reader.py", line 300, in _openStream firstRecord=firstRecordIdx) File "/home/vagrant/nupic/nupic/data/file_record_stream.py", line 164, in __init__ self._file = open(self._filename, self._mode) IOError: [Errno 2] No such file or directory: u'/home/vagrant/nupic/nupic/datafiles/rec-center-hourly.csv' I'm not sure why its looking in /home/vagrant/nupic/nupic/datafiles/rec-center-hourly.csv when the config file specifies 'file://rec-center-hourly.csv', which should mean the $PWD. I tried debugging; I set a breakpoint on line 164 of file_record_stream.py and self._filename evaluates to u'/home/vagrant/nupic/examples/opf/clients/hotgym/prediction/one_gym/rec-center-hourly.csv' which is correct. However, I think this is multithreaded code, so when the breakpoint is hit, it could be just the first time. When I continue after that breakpoint, the breakpoint isn't hit again and it fails with the above error. I also tried changing "source" in swarm_description.py to "file://extra/hotgym/rec-center-hourly.csv", and that got a little further, but then it complained that there's a discrepancy in the column names. It seems that at one time while swarm.py is run, it thinks the root is $NUPIC, and then at another time it thinks its $PWD. I'm not sure where the source of the problem is though. I could use someone's help in tracking this down. Thanks! -- Mr. Mateja Putic Ph.D Candidate Department of Electrical and Computer Engineering University of Virginia (703) 303-2099
