These are chat archives for dereneaton/ipyrad

Jul 2017
Jul 30 2017 05:59

@isaacovercast here is the info in the log file with the debug flag on: -------------------------------------------------------------
ipyrad [v.0.7.2]

Interactive assembly and analysis of RAD-seq data

Begin run: 2017-07-28 22:55
Using args {'preview': False, 'force': False, 'threads': 2, 'results': False, 'quiet': False, 'merge': None, 'ipcluster': False, 'cores': 20, 'params': 'params-3RAD_plate3Schiz_NBV.txt', 'branch': None, 'steps': '1', 'debug': True, 'new': None, 'MPI': False}
Platform info: ('Linux', 'node132', '2.6.32-504.16.2.el6.x86_64', '#1 SMP Wed Apr 22 06:48:29 UTC 2015', 'x86_64')2017-07-28 22:55:35,326 pid=23957 [] INFO ['ipcluster', 'start', '--daemonize', '--cluster-id=ipyrad-cli-23957', '--engines=Local', '--profile=default', '--n=20']
2017-07-29 01:07:17,300 pid=15236 [] INFO debugging turned on and registered to be turned off at exit

Isaac Overcast
Jul 30 2017 16:02
So it's failing in the same way for both real data and the toy tutorial data, and only on the cluster. And its printing debug, but only the header. The first line of debug it should print is inside the first jobs sent to the ipyparallel cluster, so this indicates to me that there's some kind of communication issue. You might try making sure your ipyparallel package is up to date: conda install ipyparallel -f.
Another thing to try is to fire up an interactive job and see if you can't get it working by hand on a compute node like this:
Try inside the interactive task and see if it doesn't give you a little more information.