Vakyrja is a very light deployment helper that uses rsync, ssh and a few other basic tools.

Install via,

npm i valkyrja -g

You will have access to the valk command.

If you ever need a specific copy of valkyrja for a project simply install that version into the project:

npm i -S valkyrja

After doing so, whenever you run any valk commands they will be executed by the specific version of valkyrja you have installed in the project instead of the global valkyrja

Command Help

You can setup project via valk make. For a full setup run valk make all.

You can also setup any specific files at any time via valk make user and valk make tools

Type valk for command and usage help

Here is a potentially incomplete list of available commands, more details on the command help.

make       - create required files
deploy     - deploy local copy
dry-run    - simulate deploy
diff       - diff servers with local copy
diff-file  - diff file with local copy
ls         - list servers
clean      - remove junk
check      - run tools.check
build      - run tools.build
version    - print version
help       - help page

Custom Commands

Since most deployment scenarios often involve hard to foresee extra project specific tooling we provide a very simple and easy way to add custom commands into valkyjra's main command interface. Simply add any function to your .valkyjra.js tools file and so long as the name of the function isn't used by valkyrja already your function will get called when the name is invoked on the command line. The tools file created by make has an example "test" custom command. Feel free to add entries to .valkyjra.toml and .valkyjra.user.toml to help with configuring your custom command (you will have access to the merged version as the conf parameter).

You can also add basic entries to the help screen by including a customCmds hash in your .valkyjra.js tools file; however we recommend documenting your custom commands proper in a README or similar file as such we only provide support for basic help entries with no details and do not require the hash to have an entry of the command for the custom command to work.

Files

The valk command understands the following files,

• .valkyrja.toml, main configuration file and only mandatory file. Specify servers, deploy actions, files to sync, files to ignore, etc in here; you can also pass any rsync options you like

• .valkyrja.user.toml, main user file. You can create a default one via valk make user. The contents of this file will be merged into .valkyrja.toml before any command is executed.

• .valkyrja.js is an optional tools file. valk will call the corresponding function exported by the file and expect a Promise. Based on the Promise the deploy process will continue or halt. The expectation is that you perform any production build steps or any checks and resolve or reject the promise based on the result of your tooling; deployment will continue only if promise is fulfilled.

Testing Deployment Strategy

You don't need a server.

Just point valkyrja to localhost and the path somewhere in your home directory. For running commands you will also need to install a ssh server.

For ubuntu you can install a ssh server very easily like so,

sudo apt-get install openssh-server

It's annoying but even though you're ssh'ing to your machine and the user you're already logged in as the ssh server will still request you to authenticate, to avoid this you can just add add your local ssh keys to your local ssh keys... err, it makes sense trust me.

First check you have ssh keys generated,

ls -la ~/.ssh

If you don't see id_rsa or id_rsa.pub run ssh-keygen to create them.

Now you just need to copy your key to the server, in our case your local machine, this works exactly like it would for any non-localhost server,

ssh-copy-id `whoami`@localhost

You'll be asked for your local password. You can test it worked using:

ssh `whoami`@localhost 'ls -la'

You should see it list your home directory; if you didn't do the ssh-copy-id it would ask you for the password to do that.

Useful .valkyrja.js

The generated .valkyrja.js is designed to only provide placeholders and not blow up in your face when you try to do anything. Here is an example of one that is actually useful:

var Promise = require('lie');
var cmdspawn = require('cmdspawn');
var _ = require('shadow.lodash');

var cmd = cmdspawn({
	tracking : false, // show start and end of command
	verbose  : false  // write executed commands to console
});

var excludeNPMDevDeps = function (rsync) {
	var pkg = require('./package.json');
	_.each(pkg.devDependencies, function (version, name) {
		rsync.exclude('node_modules/' + name);
	});
}

module.exports = {

	build: function (type, category, host) {
		var p = Promise.resolve();
		p = p.then(cmd.f('gulp build --color'), cmd.silent);
		return p;
	},

	check: function (type, category, host) {
		var p = Promise.resolve();

		if (type == 'all' || type == 'php') {
			p = p.then(cmd.f([
					'phpunit',
					'-c src/server/api/phpunit.xml',
					'--coverage-text',
					'--color'
				].join(' ')), cmd.silent);
		}

		if (type == 'all' || type == 'js') {
			p = p.then(cmd.f([
					'mocha',
					'--color',
					'--recursive',
					'src/client/node_modules/.spec'
				].join(' ')), cmd.silent);
		}

		return p;
	},

	confdryrun: function (conf, rsync, ssh, host) {
		excludeNPMDevDeps(rsync);
	},

	confdeploy: function (conf, rsync, ssh, host) {
		excludeNPMDevDeps(rsync);
	},

	confdiff: function (conf, rsync, ssh, host) {
		excludeNPMDevDeps(rsync);
	}

};

Gotchas

User must belong to group he is trying to change

If the user trying to sync does not belong to the group he is trying to sync to, the sync command will fail when chgrp is executed due to the command in question returning non 0 exit code.

To add a user to the group the following must be executed on the server,

sudo usermod -a -G GROUPNAME YOURUSERNAME

Important Note! make sure to have -a (ie. append) flag there or you will remove the user from every other group. If you do that to yourself on your local machine where you are in the sudo'er group you will remove yourself from the sudo'er group preventing you from executing commands with sudo

If you wish to run your own custom chgrp command you can disable the in-built one by setting autogroup = false on the root of the configuration file.

You can access the host group via the <<group>> variable when adding postdeploy commands. eg. chgrp -R -P <<group>> .

If you need extra variables to achieve the command you can add them to servers, then they will be available just like how group is available as <<group>>. But make sure they are strings and can never be anything but strings. If you add a non-string parameter to the server, it will not be available.

Ignore exit status of some commands

You may find some commands just annoyingly return non 0 for certain cases. Cases you don't really consider error cases. eg. grep returns non-0 if it can't match anything. If you have ssh.stop-on-errors this can be even more annoying then just a little bit of red text every deploy.

To ignore the exit code of a commands simply pass their output though another command. eg. if you want to ignore the exit status of a grep commands then you would write the command as grep ... | cat

Q&A

Q. What options does valkyrja pass to rsync by default?
A. Except dry-run flags, none. Only options in .valkryja.toml are passed.

Q. What options does valkyrja pass to ssh by default?
A. If ssh key is missing, timeout will be set to 10 minutes.

Q. What's a TOML file?
A. JSON with comments, for humans.

Q. What's "valkyrja" a reference to?
A. Old Norse for "chooser of the slain", aka. "valkyrie"