If a user has to manage multiple jobs in a direct acyclic graph (DAG) order it is possible to organize the submission with a DAG input file.
In the graph below, the vertices, generally tagged with a letter, represent the jobs and the edges represent the degree of relationship.
In this case the A job is a parent of the B¹, B² and B³ jobs, namely the children.
This implies that the B* jobs start once the A job finished.
This automatic implementation of successive submissions could be very useful.
To properly perform a DAG job it occurs to write a DAG input file with the specific tags of the jobs and the degree of relationship, for instance:
-bash-4.2$ cat simple.dag JOB A sleep.sub JOB B snore.sub PARENT A CHILD B |
In this case the sleep.sub job is called A, whereas the snore.sub job is B.
Moreover, A is a parent of B, so the B job starts only once the A job finished.
In order to submit properly the DAG job, the DAG input file has to be into a folder reacheable from the schedd. At tier-1, in general this folder could be a directory into the gpfs_data file system. After this, to submit the DAG job it is enough to issue the following commands
-bash-4.2$ export _condor_SCHEDD_HOST=sn-02.cr.cnaf.infn.it -bash-4.2$ condor_submit_dag simple.dag Renaming rescue DAGs newer than number 0 ----------------------------------------------------------------------- File for submitting this DAG to HTCondor : simple.dag.condor.sub Log of DAGMan debugging messages : simple.dag.dagman.out Log of HTCondor library output : simple.dag.lib.out Log of HTCondor library error messages : simple.dag.lib.err Log of the life of condor_dagman itself : simple.dag.dagman.log Submitting job(s). 1 job(s) submitted to cluster 1271844. ----------------------------------------------------------------------- |
The submission produces the log files shown in the output.
Then, to check the job status a user can launch the condor_q command
-bash-4.2$ condor_q -- Schedd: sn-02.cr.cnaf.infn.it : <131.154.192.42:9618?... @ 01/26/22 17:47:19 OWNER BATCH_NAME SUBMITTED DONE RUN IDLE TOTAL JOB_IDS arendinajuno simple.dag+1271844 1/26 16:51 _ 1 _ 1 1271846.0 Total for query: 1 jobs; 0 completed, 0 removed, 0 idle, 1 running, 0 held, 0 suspended Total for arendinajuno: 1 jobs; 0 completed, 0 removed, 0 idle, 1 running, 0 held, 0 suspended Total for all users: 26968 jobs; 18806 completed, 1 removed, 4984 idle, 3132 running, 45 held, 0 suspended |
and after the A job is done, the child job B is queued:
-- Schedd: sn-02.cr.cnaf.infn.it : <131.154.192.42:9618?... @ 01/26/22 17:48:56 OWNER BATCH_NAME SUBMITTED DONE RUN IDLE TOTAL JOB_IDS arendinajuno simple.dag+1271947 1/26 17:46 _ 1 _ 1 1271949.0 Total for query: 1 jobs; 0 completed, 0 removed, 0 idle, 1 running, 0 held, 0 suspended Total for arendinajuno: 1 jobs; 0 completed, 0 removed, 0 idle, 1 running, 0 held, 0 suspended Total for all users: 28900 jobs; 18810 completed, 1 removed, 6840 idle, 3204 running, 45 held, 0 suspended |