Dreadnot is a 'one click' deploy tool written in Node.js.
Dreadnot was heavily inspired by Etsy's Deployinator.
Dreadnot loads its configuration from a javascript file:
exports.config = {
// The name of this Dreadnot instance, used for display
name: 'Example Dreadnot',
// Each Dreadnot instance supports one environment such as 'dev', 'staging'
// or 'production'
env: 'staging',
// The data root Dreadnot will use
data_root: '/var/dreadnot',
// Base URL to access dreadnot (used in IRC, email, Hipchat)
default_url: 'http://example.com',
// Dreadnot uses an htpasswd file (with support for bcrypt, md5 and sha1) for auth
htpasswd_file: '/etc/dreadnot/htpasswd',
// Each stack represents a code base that should be deployed to one or more regions
stacks: {
// For a stack named 'webapp', there should be a 'webapp.js' file in the
// stacks directory
webapp: {
// What branch to look in for the latest revision of the code base
tip: 'master',
// How long to cache the latest revision of the code base
tip_ttl: 120 * 1000,
// What regions this stack should be deployed to
regions: ['ord1'],
// Stacks should implement dryrun for testing
dryrun: true
}
},
// The GitHub organization you provide is used to build URLs for your stacks
github: {
organization: 'racker'
},
// Plugins provide optional functionality such as notifications. Any plugins
// that are not configured won't be used.
plugins: {
// An IRC notification plugin
irc: {
nick: 'staging-dreadnot',
channels: {'irc.freenode.net': ['#public-channel', '#private-channel pass']}
},
// An email notification plugin
email: {
server: {
user: '[email protected]',
password: '',
host: 'smtp.example.com',
ssl: true
},
to: '[email protected]',
from: '[email protected]'
},
// A Hipchat notification plugin
hipchat: {
name: 'Dreadnot',
apiKey: '123456789abcdefg',
rooms: [
1234,
5678
]
}
}
};Dreadnot looks in a directory (by default ./stacks, but this can be changed
from the command line) for "stack files". A stack file is simply a javascript
file that exports
- A
get_deployedRevisionfunction which takes an object containingenvironmentandregionfields, and a callback taking(err, deployed_revision). - A
targetshash that maps target names to lists of task names. Currently, the only supported targets aredeploy, which defaults to['task_preDeploy', 'task_deploy', 'task_postDeploy'], andfinallywhich does not have a default value. You should use thefinallytarget if there are any tasks you would like to run every time, regardless of the success or failure of the tasks indeploy(i.e. re-enable monitoring alerts). The tasks in thefinallytarget itself are each dependent on the success of the last task in the target, so an error in one will prevent the rest from running. - One or more "task functions" whose names are prefixed with
task_. Each task function takes:- A "stack" object. The most useful fields on the stack are
stackConfigwhich contains the config for this particular stack, andconfigwhich contains the global config. - A "baton" object. Each task executed during a run of a given target
receives the same baton object. By default, it contains a
logfield with methods such asdebug,info, anderrorthat can be used to log output to deployment log and web view. - An "args" hash with
dryrun,environment,region,revisionanduser, each of which is a string. - A "callback" function that should be called without arguments on completion, or with a single error object if an error occurs.
- A "stack" object. The most useful fields on the stack are
In the configuration used by Rackspace Cloud Monitoring, a deployment looks something like:
- Build: verify that the requested revision has been successfully built and that all tests pass.
- Prepare: remove the region being deployed from the load balancer rotation, redirecting all traffic to another region.
- Execute: use a chef search to locate all servers in the region, then ssh to each in parallel to upgrade the code.
- Validate: execute checks against each upgraded service to verify that it is functioning properly.
- Restore: restore the region to the load balancer rotation.
Imporantly, Dreadnot knows nothing about the hosts to which it is deploying - if it did, we would have to modify our Dreadnot configuration every time we added or removed a machine from our environment. Instead, we rely on chef (although anything that knows about your servers will work) to give us an up-to-date list of all hosts in a given region. In smaller deployments it might be suitable to hardcode a list of hosts.
Does Dreadnot support SVN?
Dreadnot supports Node.js - you can use any technology or topology that suits you, as long as you can find a library for it.
To create a development environment, you'll need Vagrant and Virtualbox. Once installed, run:
vagrant up
Then visit http://localhost:8000
Log into the VM by running and running common commands:
vagrant ssh
sudo cat /var/log/upstart/dreadnot.log
npm install dreadnot -g
Alternatively, when developing, you can find a compiled dreadnot binary in the bin folder.
Dreadnot takes a number of options on the command line. The defaults are:
dreadnot -c ./local_settings.js -s ./stacks -p 8000
This will start dreadnot with the specified config file and stack directories, listening on port 8000.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent