You can use 'pq' to queue jobs on your machine, i.e. execute programs depending on your current system load. Recognized limits are the current load-average value (as given, e.g. by 'uptime' or 'top') as well as the current total memory usage (as given, e.g. by 'free'). If the limits -- which can be controlled by the '--sys-load-limit' and '--sys-mem-limit' options -- for the given values have not been reached, the system's state is interpreted as idle and the next job in the queue is run.
pq does not include a daemon. Job execution is triggered only by execution of 'pq -u', resp. 'pq --update'. To automatically process the queue, run a separate script to regularly call 'pq -u', e.g. by defining a cron-job (via 'crontab -e'). A typical cron-job definition to update the queue every two minutes would e.g. be '*/2 * * * * /path/to/pq --update'.
pq is run on a per-user basis and different pq-instances do not communicate with each other. The only way pq decides whether to run the next job or not is by watching the system usage. The pq configuration is located at $HOME/.config/pq. This directory contains pq.conf, the configuration file and other data like the jobs-database.
A job is easily added by
pq "some_cmd --an-option --other-option args"
Everything in quotes will be interpreted as an ordinary shell command and be executed as such. pq automatically saves the directory from which the job has been submitted and will execute the command in that directory. Additionally, pq will save current environment variables and set them before job execution in the executing shell. The environment variables to be preserved can be set in the pq.conf-file, per default pq will save $HOME, $LD_LIBRARY_PATH, $PATH and $PYTHONPATH.
Report bugs to sittel@lettis.net.