README in whiskey_disk-0.0.7 vs README in whiskey_disk-0.3.0

- old
+ new

@@ -1,41 +1,47 @@ Whiskey Disk -- embarrassingly fast deployments. - A very opinionated deployment tool, designed to be as fast as technologically - possible. (For more background, read the WHY.txt file) + A very opinionated deployment tool, designed to be as fast as technologically possible. (For more background, read the WHY.txt file) - Should work with any project which is git hosted, not just Ruby / Ruby on Rails projects. - Allows for local deploys as well as remote. + Should work with any project which is git hosted, not just Ruby / Ruby on +Rails projects. Allows for local deploys as well as remote. Selling points: - - If you share the same opinions there's almost no code involved, almost no dependencies, - and it uses stock *nix tools (ssh, bash, rsync) to get everything done. + - If you share the same opinions there's almost no code involved, almost no + dependencies, and it uses stock *nix tools (ssh, bash, rsync) to get +everything done. - - Written completely spec-first for 100% coverage. We even did that for the rake tasks, - the init.rb and the plugin install.rb if you swing that way. + - Written completely spec-first for 100% coverage. We even did that for the + rake tasks, the init.rb and the plugin install.rb if you swing that way. - - 1 ssh connection per rake task -- so everything needed to do a full setup is done in - one shot. Everything needed to do a full deployment is done in one shot. Having 8 - minute deploys failing because I was on cdma wireless on a train in india where the - connection won't stay up for more than 2-3 minutes is not where I want to be. + - 1 ssh connection per rake task -- so everything needed to do a full setup + is done in one shot. Everything needed to do a full deployment is done in +one shot. Having 8 minute deploys failing because I was on cdma wireless on a +train in india where the connection won't stay up for more than 2-3 minutes is +not where I want to be. - - You can do *local* deployments, by this I mean you can use whiskey_disk to manage your - local checkout with localized (aka non-production) configurations trivially. This turns - out to be surprisingly handy (well, I was surprised). + - You can do *local* deployments, by this I mean you can use whiskey_disk to + deploy fully running instances of your application to the same machine +you're developing on. This turns out to be surprisingly handy (well, I was +surprised). *NOTE*: be sure to set your deploy_to to a place other than the +current local checkout. - - You can have per-developer configurations for environments (especially useful for - "local" or "development" enviroments). Use .gitignore and everyone can have their own - local setup that just works. + - You can have per-developer configurations for environments (especially + useful for "local" or "development" enviroments). Use .gitignore, or +specify a config_branch and everyone can have their own local setup that just +works. - - Deployment configuration is specified as YAML data, not as code. Operations to perform - after setup or deployment are specified as rake tasks. + - Deployment configuration is specified as YAML data, not as code. + Operations to perform after setup or deployment are specified as rake +tasks. - - There's no before_after_before_after hooks. You've got plenty of flexibility with just - a handful of rake hook points to grab onto. + - There's no before_after_before_after hooks. You've got plenty of + flexibility with just a handful of rake hook points to grab onto. - - rake deploy:setup and rake deploy:now are really all that are needed, best I can tell. + - rake deploy:setup and rake deploy:now are really all that are needed, best + I can tell. Dependencies: rake, ssh, git, rsync on the deployment target server (affectionately referred to as the "g-node" by vinbarnes), bash-ish shell on deployment @@ -61,107 +67,148 @@ Configuration: - look in the examples/ directory for sample configuration files - main configuration is in <app_root>/config/deploy.yml - - per-environment configurations can be stored in <app_root>/config/deploy-<environment>.yml - config files are YAML, with a section for each environment. Known config file settings (if you're familiar with capistrano and vlad these should seem eerily familiar):: - domain: host on which to deploy (this is an ssh connect string) - deploy_to: path to which to deploy main application - repository: git repo path for main application - - branch: git branch to deploy from main application git repo + - branch: git branch to deploy from main application git repo (default: master) - deploy_config_to: where to deploy the configuration repository - config_repository: git repository for configuration files + - config_branch: git branch to deploy from configuration git repo (default: master) - project: project name (used to compute path in configuration checkout) - rake_env: hash of environment variables to set when running post_setup and post_deploy rake tasks - - defining a deploy:<environment>:post_setup rake task (e.g., in lib/tasks/ or in - your project's Rakefile) will cause that task to be run at the end of deploy:setup + - defining a deploy:<environment>:post_setup rake task (e.g., in lib/tasks/ + or in your project's Rakefile) will cause that task to be run at the end +of deploy:setup - - defining a deploy:<environment>:post_deploy rake task (e.g., in lib/tasks/ or in - your project's Rakefile) will cause that task to be run at the end of deploy:now + - defining a deploy:<environment>:post_deploy rake task (e.g., in + lib/tasks/ or in your project's Rakefile) will cause that task to be run +at the end of deploy:now - Running: - - rake deploy:setup to=<environment> (e.g., "staging", "production", etc.) - - rake deploy:now to=<environment> + Running via rake: + ... in your Rakefile: + require 'whiskey_disk/rake' + + Then, from the command-line: + + % rake deploy:setup to=<target> (e.g., "staging", "production", etc.) + % rake deploy:now to=<target> + + or, specifying the project name: + + % rake deploy:setup to=<project>:<target> (e.g., "foo:staging", "bar:production", etc.) + % rake deploy:now to=<project>:<target> + + + Running from the command-line: + + % wd setup --to=<target> + % wd setup --to=<project>:<target> + % wd setup --to=foo:staging --path=/etc/whiskey_disk/deploy.yml + + % wd deploy --to=<target> + % wd deploy --to=<project>:<target> + % wd deploy --to=foo:staging --path=/etc/whiskey_disk/deploy.yml + + + Note that the wd command (unlike rake, which requires a Rakefile in the current directory) can be run from anywhere, so you can deploy any project, working from any path, and can even specify where to find the deployment YAML configuration file. + + The --path argument can take either a file or a directory. When given a file it will use that file as the configuration file. When given a directory it will look in that directory for deploy/<project>/<target>.yml, then deploy/<project>.yml, then deploy/<target>.yml, then <target>.yml, and finally, deploy.yml. + + All this means you can manage a large number of project deployments (local or remote) and have a single scripted deployment manager that keeps them up to date. Configurations can live in a centralized location, and developers don't have to be actively involved in ensuring code gets shipped up to a server. Win. + + Configuration Repository What's all this about a second repository for configuration stuff? - Well, we're going to make it optional probably in another release, but we really - are digging this, so maybe you should try it. Basically it goes like this... + This is completely optional, but we really are digging this, so maybe +you should try it. Basically it goes like this... - We have a number of web applications that we manage. Usually there's a customer, - there might be third-party developers, or the customer might have access to the git - repo, or their designer might, etc. We also tend to run a few instances of any given - app, for any given customer. So, we'll run a "production" site, which is the public- - facing, world-accessible main site. We'll usually also run a "staging" site, which is - roughly the same code, maybe the same data, running on a different URL, which the customer - can look at to see if the functionality there is suitable for deploying out to production. - We sometimes run a "development" site which is even less likely to be the same code as - production, etc., but gives visibility into what might end up in production one day soon. + We have a number of web applications that we manage. Usually there's a +customer, there might be third-party developers, or the customer might have +access to the git repo, or their designer might, etc. We also tend to run a +few instances of any given app, for any given customer. So, we'll run a +"production" site, which is the public- facing, world-accessible main site. +We'll usually also run a "staging" site, which is roughly the same code, maybe +the same data, running on a different URL, which the customer can look at to +see if the functionality there is suitable for deploying out to production. We +sometimes run a "development" site which is even less likely to be the same +code as production, etc., but gives visibility into what might end up in +production one day soon. - So we'll store the code for all of these versions of a site in the same git repo, typically - using a different remote branch for each environment ("production", "staging", "development"). + So we'll store the code for all of these versions of a site in the same git +repo, typically using a different remote branch for each environment +("production", "staging", "development"). - One thing that comes up pretty quickly is that there are various files associated with - the application which have more to do with configuration of a running instance than they - have to do with the application in general. In the rails world these files are probably - in config, or config/initializers/. Think database connection information, search engine - settings, exception notification plugin data, email configuration, Amazon S3 credentials, - e-commerce back-end configuration, etc. + One thing that comes up pretty quickly is that there are various files +associated with the application which have more to do with configuration of a +running instance than they have to do with the application in general. In the +rails world these files are probably in config, or config/initializers/. Think +database connection information, search engine settings, exception notification +plugin data, email configuration, Amazon S3 credentials, e-commerce back-end +configuration, etc. - We don't want the production site using the same database as the development site. We don't - want staging using (and re-indexing, re-starting, etc.) production's search engine server. - We don't want any site other than production to send account reset emails, or to push orders - out to fulfillment, etc. + We don't want the production site using the same database as the +development site. We don't want staging using (and re-indexing, re-starting, +etc.) production's search engine server. We don't want any site other than +production to send account reset emails, or to push orders out to fulfillment, +etc. - For some reason, the answer to this with cap and/or vlad has been to have recipes which - reference various files up in a shared area on the server, do copying or symlinking, etc. - Where did those files come from? How did they get there? How are they managed over time? - If they got there via a configuration tool, why (a) are they not in the right place, or - (b) do we have to do work to get them into the right place? + For some reason, the answer to this with cap and/or vlad has been to have +recipes which reference various files up in a shared area on the server, do +copying or symlinking, etc. Where did those files come from? How did they get +there? How are they managed over time? If they got there via a configuration +tool, why (a) are they not in the right place, or (b) do we have to do work to +get them into the right place? - So, we decided that we'd change how we deal with the issue. Instead of moving files around - or symlinking every time we deploy, we will manage the configuration data just like we - manage other files required by our projects -- with git. + So, we decided that we'd change how we deal with the issue. Instead of +moving files around or symlinking every time we deploy, we will manage the +configuration data just like we manage other files required by our projects -- +with git. - So, each project we deploy is associated with a config repo in our git repository. Usually - many projects are in the same repo, because we're the only people to see the data and - there's no confidentiality issue. But, if a customer has access to their git information - then we'll make a separate config repo for all that customers' projects. (This is easier - to manage than it sounds if you're using gitosis, btw.) + So, each project we deploy is associated with a config repo in our git +repository. Usually many projects are in the same repo, because we're the only +people to see the data and there's no confidentiality issue. But, if a +customer has access to their git information then we'll make a separate config +repo for all that customers' projects. (This is easier to manage than it +sounds if you're using gitosis, btw.) - Anyway, a config repo is just a git repo. In it are directories for every project - whose configuration information is managed in that repo. For example, there's a "larry" - directory in our main config repo, because we're deploying the larry project - (http://github.com/rick/larry) to manage our high-level configuration data. + Anyway, a config repo is just a git repo. In it are directories for every +project whose configuration information is managed in that repo. For example, +there's a "larry" directory in our main config repo, because we're deploying +the larry project (http://github.com/rick/larry) to manage our high-level +configuration data. - Note, if you set the 'project' setting in deploy.yml, that determines the name of the - top-level project directory whiskey_disk will hunt for in your config repo. If you don't - it uses the 'repository' setting (i.e., the git URL) to try to guess what the project - name might be. So if the URL ends in foo/bar.git, or foo:bar.git, or /bar, or :bar, - whiskey_disk is going to guess "bar". If it's all bitched up, just set 'project' manually - in deploy.yml. + Note, if you set the 'project' setting in deploy.yml, that determines the +name of the top-level project directory whiskey_disk will hunt for in your +config repo. If you don't it uses the 'repository' setting (i.e., the git URL) +to try to guess what the project name might be. So if the URL ends in +foo/bar.git, or foo:bar.git, or /bar, or :bar, whiskey_disk is going to guess +"bar". If it's all bitched up, just set 'project' manually in deploy.yml. - Inside the project directory is a directory named for each environment we might - deploy to. Frankly, we've been using "production", "staging", "development", and "local" - on just about everything. + Inside the project directory is a directory named for each environment we +might deploy to. Frankly, we've been using "production", "staging", +"development", and "local" on just about everything. - Inside the environment directory is a tree of files. So, e.g., there's config/, which - has initializers/ and database.yml in it. + Inside the environment directory is a tree of files. So, e.g., there's +config/, which has initializers/ and database.yml in it. - Long story short, load up whatever configuration files you're using into the repo - as described, and come deployment time exactly those files will be overlaid on top of - the most recent checkout of the project. Snap. + Long story short, load up whatever configuration files you're using into +the repo as described, and come deployment time exactly those files will be +overlaid on top of the most recent checkout of the project. Snap. project-config/ | +---larry/ | @@ -199,31 +246,28 @@ - We are using this to manage larry. See: http://github.com/rick/larry/blob/master/config/deploy.yml http://github.com/rick/larry/blob/master/lib/tasks/deploy.rake - Note that you can also provide per-environment config settings, - outside of deploy.yml. This is most useful for handling per-developer - local deployments, though you could use it to override some settings - if that makes sense in some context. Here's a simple example of that: + Note that you can also provide per-environment config settings, outside of +deploy.yml. This is most useful for handling per-developer local deployments, +though you could use it to override some settings if that makes sense in some +context. Here's a simple example of that: http://github.com/rick/larry/blob/master/config/deploy-local.yml.example - We are using it on a private project with lots of config files, but here's a gist showing a bit more interesting deploy.rake file for post_setup and - post_deploy work: +post_deploy work: https://gist.github.com/47e23f2980943531beeb - - On the radar for an upcoming release are: - common post_* recipes being specifiable by symbol names in deploy.yml ? - - making the config repo stuff optional if you don't need that at all - - bringing in a very simplified version of mislav/git-deploy post-{receive,reset} - hooks, and a little sugar so you can say: + - bringing in a very simplified version of mislav/git-deploy + post-{receive,reset} hooks, and a little sugar so you can say: task :post_deploy do if has_changes?('db/migrate') or has_changes?('config/database.yml') Rake::Task['db:migrate'].invoke end