Go to file

R Tyler Croy 97ae0d52a6 Add more sketches		2020-12-30 09:43:02 -08:00
cli	Add some sketches of what tasks and plans might look like	2020-12-29 22:16:52 -08:00
.gitignore	Initial commit	2020-12-28 16:46:39 -08:00
Cargo.lock	Play around with simple SSH commands	2020-12-28 22:12:45 -08:00
Cargo.toml	Play around with simple SSH commands	2020-12-28 22:12:45 -08:00
README.adoc	Add more sketches	2020-12-30 09:43:02 -08:00
inventory.yml	Allow running commands across a group	2020-12-28 22:34:48 -08:00

README.adoc

Zap

A simple cross-platform orchestration and configuration management tool.

The main goal for Zap is to a simple mechanism for managing groups of computers with varying configurations and needs. Zap accomplishes this with "tasks" which can be composed into "plans" or run standalone. These tasks can be collections of scripts or a statically linked binaries, which will be pushed to the target machine(s) and executed.

Zap borrows ideas from Puppet Bolt. but leaves some of the Puppet-based legacy from Bolt behind.

Goals

Support BSD and Linux with ease
Make creating a plan very easy, including adding some simple logic
Explore tags-based task assignment
Pulll dependencies in through git-subtree or other, no inventing a new package distribution mechanism

Sketches

install.ztask

/*
 * The FileSync should be an internal task type
 */
task FileSync {
    parameters {
        localfile {
            type = string
            help = "Path on the local system for the file to sync"
            required = true
        }
        remotefile {
            type = "string"
            help = "Path on the remote system for the file"
            required = true
        }
    }
}

// Sketch of a user-defined  task
task Install {
    // The restrict block is a collection of expressions which ensure that the
    // task is not run anywhere it cannot run. Will fail if it's tried to run
    // somewhere it cannot be run
    restrict {
        // Regular expression to define what
        match_fact("platform", "(.*)-freebsd")
        match_fact("hostname", "www1")
    }

    parameters {
        package {
            type = string
            help = 'The package to be installed'
            required = true
        }

        unless {
            type = string
            help = "Script which when returns zero if the package has been installed, i.e. `test -f /usr/bin/nginx`"
        }
    }

    // Parameters exposed as environment variables
    // including the "ZAP_NOOP" variable which will be set if the script should
    // be run in noop
    script {
        // either the file or the inline must be defined
        file = "path/to/file/in/tree"
        // When using `program`, the task should expect to find:
        // command-name_x86_64-unknown-linux-gnu
        program = "/path/to/exes/in/tree"
        inline = """
        """
    }
}


/ Exploring module syntax for namespacing these */
module Install {
    task Pkg {
        restrict {}
        parameters {}
        script {}
    }

    task Zypper {
        restrict {}
        parameters {}
        script {}
    }
}

Sketch of a zplan

// Should things be just done proceduraly?

task("sync-tree.ztask") { }

task('tasks/install.ztask') {
  name = "install-nginx"
  package = "nginx"
  unless = "test -f /usr/bin/nginx"
}

// Run another plan from within this plan (supdawg)
plan("plans/prepare-website.zplan") {
  // What parameters make sense here?
}


// Relationships between tasks, option A
task('tasks/sync-tree.zplan') {
    source = './foo'
    destination = '/'

    then = ["install-nginx"]
}

// Relationships between tasks, option B
task('tasks/sync-tree.zplan') {
    source = './foo'
    destination = '/'

    then {
        task("tasks/install.ztask") {
            package = "nginx"
            unless = "test -f /usr/bin/nginx"
        }
    }
}

Last updated 2024-05-18 17:05:01 UTC

</html>