qstat: Meta-data for currently running jobs

In patr1ckm/distributr: Grid Search with SGE Backend

Description

The meta-data of running SGE jobs are returned as a data frame. The meta includes status, resources requested, memory usage, and wallclock/cpu time. Works for all grid engine jobs and task arrays, including the ones created by distributr.

Usage

qstat(user = TRUE)

Arguments

user	job meta-data is returned for user jobs only (default: TRUE), otherwise meta-data for all jobs in default queue is returned

Value

A data frame with the following columns:

job_id	SGE job id
.sge_id	the value of SGE_TASK_ID for the job
prior	Priority of the job
name	job name
user	user name
state	state of the job, e.g. r for running
start	date and time of job start
queue	queue and machine job is running on
jclass	class of job (usually NA)
slots	Number of slots (cores) job is running with
maxvmem	Maximum virtual memory used for the job (GB)
mem	Current physical memory used for the job (GB)
vmem	Current virtual memory used for the job (GB)
wallclock	Amount of time the job has been running (sec)
cpu	Amount of CPU time the job has used (sec)

The columns prior, user, start, queue, and jclass are not printed. If no jobs are running or in the queue, returns data.frame(). For queued jobs, queue = NA, and .sge_id is the lowest task still in the queue. Does not work for jobs with job classes.

Examples

## Not run: 

sge_test(wait=20)
qstat()

## End(Not run)

patr1ckm/distributr documentation built on May 24, 2019, 8:21 p.m.