cooperate 0.3.1

Distribute commands to many nodes

cooperate is a shell command that execute commands in a cooperative manner, by distributing them to many nodes.

It requires Python >= 3.3 and asyncio.

For example:

cooperate --local -- echo FOO

Will execute the job “echo FOO” locally.

This one:

cooperate --ssh [e-mail] -- echo FOO

Is barelly equivalent to:

ssh [e-mail] echo FOO

You can declare as many nodes as you want. For example:

cooperate --local --ssh [e-mail] --ssh [e-mail] -- echo FOO

Is equivalent to:

echo FOO
ssh [e-mail] echo FOO
ssh [e-mail] echo FOO

You can also declare many jobs at once. For example:

cooperate --local --command "echo FOO" --command "echo BAR"

Is equivalent to:

echo FOO
echo BAR

Installation

pip install cooperate

Nodes

Commands can be distribued thru these kind of nodes:

• –local execute locally

• –ssh execute thru ssh

• –docker execute in a local docker container

• –lxc execute in a local lxc container

These options can be repeated as often as needed.

Modes

By default, it spawns every commands to every nodes.

The -m, –mode allow to configure the desired mode.

The all mode executes all commands in all nodes:

cooperate --local --ssh [e-mail] --ssh [e-mail] \
    --command="echo FOO" --command="echo BAR"

Is equivalent to:

echo FOO
echo BAR
ssh [e-mail] echo FOO
ssh [e-mail] echo BAR

The distribute mode share the commands out among all the nodes:

cooperate --local --ssh [e-mail] --ssh [e-mail] \
    --command="echo FOO" --command="echo BAR" --mode=distribute

Is equivalent to:

echo FOO
ssh [e-mail] echo BAR

Concurrency

By default, it executes all jobs simultaneously.

The -b, –batch option allows to execute on only a specify number of jobs at a time. Both percentages and finite numbers are supported:

cooperate --local --concurrence 1 \
    --command="echo FOO" --command="echo BAR" --command="echo BAZ"

cooperate --local --concurrence 33% \
    --command="echo FOO" --command="echo BAR" --command="echo BAZ"

The concurrency system maintains a window of running jobs. When a job returns then it starts a remnant job and so on.