Repository navigation
Job is waiting/stucked after task 3567 (collins) #387
Description
Guys,
My jobs have been waiting quite long time for next task after completing task 3561, does task 3567 takes long time to run or this is file lock issue or i am doing something wrong?
Task which have finished/ completed:
vishalnayanshi@ubuntu:~/.openml/cache/tasks$ ls
11 15 20 21 23 29 3022 3481 3493 3512 3560 3573 41 49 6
12 16 2074 219 24 3 31 3485 3494 3543 3561 36 43 53
14 18 2079 22 28 3021 32 3492 3510 3549 3567 37 45 58
Task already finished: 3561 (/home/vishalnayanshi/experiments/defaults/random_forest/vanilla/3/random_forest/vanilla/6/random_forest/vanilla/11/random_forest/vanilla/12/random_forest/vanilla/14/random_forest/vanilla/15/random_forest/vanilla/16/random_forest/vanilla/18/random_forest/vanilla/20/random_forest/vanilla/21/random_forest/vanilla/22/random_forest/vanilla/23/random_forest/vanilla/24/random_forest/vanilla/28/random_forest/vanilla/29/random_forest/vanilla/31/random_forest/vanilla/32/random_forest/vanilla/36/random_forest/vanilla/37/random_forest/vanilla/41/random_forest/vanilla/43/random_forest/vanilla/45/random_forest/vanilla/49/random_forest/vanilla/53/random_forest/vanilla/58/random_forest/vanilla/219/random_forest/vanilla/2074/random_forest/vanilla/2079/random_forest/vanilla/3021/random_forest/vanilla/3022/random_forest/vanilla/3481/random_forest/vanilla/3485/random_forest/vanilla/3492/random_forest/vanilla/3493/random_forest/vanilla/3494/random_forest/vanilla/3510/random_forest/vanilla/3512/random_forest/vanilla/3543/random_forest/vanilla/3549/random_forest/vanilla/3560/random_forest/vanilla/3561/run.xml)
[2017-12-21 15:40:20] Obtained task 3567 (collins); 24.0 attributes; 500.0 observations
Pipeline(steps=[('imputation', ConditionalImputer(axis=0, categorical_features=[19, 20], copy=True,
fill_empty=0, missing_values='NaN', strategy='median',
strategy_nominal='most_frequent', verbose=0)), ('hotencoding', OneHotEncoder(categorical_features=[19, 20], dtype=<class 'numpy.float6...estimators=10, n_jobs=1, oob_score=False, random_state=1,
verbose=0, warm_start=False))])
{}
Task already finished: 3567 (/home/vishalnayanshi/experiments/defaults/random_forest/vanilla/3/random_forest/vanilla/6/random_forest/vanilla/11/random_forest/vanilla/12/random_forest/vanilla/14/random_forest/vanilla/15/random_forest/vanilla/16/random_forest/vanilla/18/random_forest/vanilla/20/random_forest/vanilla/21/random_forest/vanilla/22/random_forest/vanilla/23/random_forest/vanilla/24/random_forest/vanilla/28/random_forest/vanilla/29/random_forest/vanilla/31/random_forest/vanilla/32/random_forest/vanilla/36/random_forest/vanilla/37/random_forest/vanilla/41/random_forest/vanilla/43/random_forest/vanilla/45/random_forest/vanilla/49/random_forest/vanilla/53/random_forest/vanilla/58/random_forest/vanilla/219/random_forest/vanilla/2074/random_forest/vanilla/2079/random_forest/vanilla/3021/random_forest/vanilla/3022/random_forest/vanilla/3481/random_forest/vanilla/3485/random_forest/vanilla/3492/random_forest/vanilla/3493/random_forest/vanilla/3494/random_forest/vanilla/3510/random_forest/vanilla/3512/random_forest/vanilla/3543/random_forest/vanilla/3549/random_forest/vanilla/3560/random_forest/vanilla/3561/random_forest/vanilla/3567/run.xml)
Activity
joaquinvanschoren commentedon Dec 21, 2017
@janvanrijn @mfeurer There is definitely something wrong with those run.xml paths, right?
@vchouksey You can speed up experiments by setting n_jobs=-1 but I'm more worried about those paths.
janvanrijn commentedon Dec 21, 2017
Can you give a MWE of your code?
vchouksey commentedon Dec 21, 2017
path have been creating with below for loop:
output_dir=os.path.expanduser('~') + '/experiments/defaults'
output_save_folder_suffix = '/vanilla'
for task_id in study.tasks:
task = openml.tasks.get_task(task_id)
output_dir = output_dir + '/' + classifier + output_save_folder_suffix + '/' + str(task_id)
I am still following similar steps of your openmlpimp library program https://github.com/janvanrijn/openml-pimp/blob/master/openmlpimp/generatedata/run_default.py
only difference is i made code in jupyter notebook each line in new cell and setting up below values rather args parser method
all_classifiers = ['random_forest']
output_dir=os.path.expanduser('~') + '/experiments/defaults'
study_id=14
openml_apikey='6beed8d7f35c545f51ba389ba80c24d3'
classifier='random_forest'
fixed_parameters = None
mfeurer commentedon Jan 4, 2018
@vchouksey what OS are you using?
vchouksey commentedon Jan 4, 2018
ubuntu 16.04 LTS
vchouksey commentedon Jan 5, 2018
same issue still happening is that big task id 3481
vchouksey commentedon Jan 5, 2018
My job has been getting killed automatically, see screenprint
vchouksey commentedon Jan 11, 2018
This issue can be close, This issue was happening because of small RAM(1 GB) of my virtual environment when I increase to 6 GB it perfectly runs without any issue, Now loads of data is already generated on Openml, Hence I would request to close this issue
joaquinvanschoren commentedon Jan 11, 2018
Good.
But what about those weird paths? Is that a bug in openmlpimp?
mfeurer commentedon Jan 12, 2018
@joaquinvanschoren please open an issue in the OpenML PIMP issue tracker to discuss paths in OpenML PIMP.
joaquinvanschoren commentedon Jan 12, 2018
OK, if you're sure it's not a openml-python bug :)
mfeurer commentedon Jan 12, 2018
openml-python does not store runs if that's actually the issue.