cwb_condor¶
This page provides a description of the cwb_condor command.
Usage
This command prepares and submits the analysis jobs on the considered computing cluster by using the condor batch system.
Syntax
cwb_condor
(without arguments)- Prints help
cwb_condor benchmark
- Prints help in the benchmark type
cwb_condor action [dag_file/cwb_stage] [input_dir]
- Prepares and submits the jobs
Further informations
The following options can be passed to cwb_condor:
action = create : it creates the dag and sub files under the condor directory submit : it submits the jobs to the considered computing cluster recovery : it compares the list of jobs in the dag file and checks the number of jobs completed (from history). It produces the dag file data_label.dag.recovery.x (x = recovery version). resume : same as recovery but is done only if previous cwb_stage is present in the output dir status : it shows the jobs status. remove : (R/H/I) it remove the jobs with status (Running/Held/Idle). check : it checks which jobs have been completed. (by reading the history reported in the output root files). list : it lists the jobs reported in the dag file. benchmark : it shows the computation load and related statistics (see cwb_condor benchmark). cleanup : remove broken symbolic links in the condor log dir (avoid init condor failure) to be used when jobs are in held status. sdag : read the standard \*.dag file produced with cwb_condor create and produced a new dag file \*.sdag The \*.sdag permits to submit in each node N jobs in sequential mode mtpe : used to generate the multitask mode for the Parameter Estimation analysis (see How to do the Parameter Estimation Analysis).dag_file/cwb_stage (optional) = dag_file : path to the dag file dag_file to be submitted (used as cwb_condor submit dag_file). cwb_stage : used in the 2G analysis [FULL(default)/INIT/STRAIN/CSTRAIN/COHERENCE/SUPERCLUSTER/LIKELIHOOD]input_dir (optional) : it is used with the recovery option. The default input directory is the output_dir directory defined in cwb_parameters.C
Examples
- The following command lines launch a full stage analysis:
cwb_condor create
cwb_condor submit
- The following command lines launch a two-stage analysis:
cwb_condor create SUPERCLUSTER
cwb_condor submit
cwb_condor recovery LIKELIHOOD outpu
cwb_condor submit condor/XXX.dag.recovery.x
- The following command line creates the condor/data_label.sub and condor/data_label.dag condor files:
cwb_condor create
- The following command lines submit the analysis jobs to the considered computing cluster:
cwb_condor submit
cwb_condor submit condor_data/data_label.dag
- The following command line recovers the analysis which did not get completed and creates the file condor/data_label.dag.recovery.1:
cwb_condor recovery
- The following command line submits the dag files listing the recovered files:
cwb_condor submit condor/data_label.dag.recovery.1
- The following command line creates the condor files for the SUPERCLUSTER stage of the 2G analysis:
cwb_condor create SUPERCLUSTER
The intermediate jobs are stored in the node’s temporary directory. Symbolic links to the remote job files are created in the output directory.
- The following command lines complete an analysis conducted with the 2G cWB pipeline starting from an intermediate stage:
cwb_condor recovery LIKELIHOOD
cwb_condor submit condor/data_label.dag.recovery.1
- The following command line creates the condor/file.sdag file from the condor/file.dag (it defines 10 sequential jobs per node):
cwb_condor sdag 10 condor/file.dag