What is the simplest queue job manager for linux?

Question

Frequently, I and my team of 6 members submit tasks to a computer. We want that the computer loads the next job after the previous one is done. We all use the same user. In another computer we used TORQUE. We also used SLURM. But they seems to be a complicated tool to configure. It is designed for multiple clusters. We have only one computer. Is there a simpler alternative? The operative system that we use is Linux (ubuntu)

Answer 1

The batch command can be used as a minimally-features scheduling system for single-user single-node cases.

You submit a job through the standard input of batch:

echo sleep 300 | batch

or you can run batch and type the commands in the batch prompt.

You monitor the queue with

atq

and inspect jobs with

at -c <jobid>

The system will start jobs one by one when the CPU load of the machine falls below a certain threshold.

For more information, refer to the man page. Note that you might have to install the at package and start the atd daemon.

If that is too minimal for your need, note that the venerable task-spooler project was revived and offers more feature, although still less than Slurm, and has both a CPU and GPU version.