"Why'd you name this repo dk?" "Don't know"
This is some automated task runner thingy ala cap/rake (except without all the drama, breaking changes and difficult-to-test-ness). You define tasks using classes; these tasks do stuff (maybe run some local or remote system commands?); you write tests for these tasks; you run them with a CLI.
First define a task:
require 'dk/task'
class MyTask
include Dk::Task
desc "my task that does something great"
def run!
log_info "this task does something"
# ... do something ...
end
end
Now route this task with a name so it can be run from the CLI:
# in config/dk.rb or whatever
require 'dk'
require 'my_task'
Dk.configure do
task 'my-task', MyTask
end
Now run this task using the CLI:
$ dk -T
my-task # my task that does something great
$ dk my-task
See dk-abdeploy for another example using Dk tasks to deploy code.
$ dk --help
Usage: dk [TASKS] [options]
Tasks:
my-other-task # my other task that does something great
my-task # my task that does something great
Options:
-T, --[no-]list-tasks list all tasks available to run
-d, --[no-]dry-run run the tasks without executing any local/remote cmds
-t, --[no-]tree print out the tree of tasks/sub-tasks that would be run
-v, --[no-]verbose run tasks showing verbose (ie debug log level) details
--version
--help
Use this option (or its -T
abbrev) to list out all tasks that are available to run. This is similar to the -T
option on cap and rake.
$ dk -T
my-other-task # my other task that does something great
my-task # my task that does something great
This option runs the tasks and logs everything just like the live runner does. However, this runner disables all system commands: both local commands and remote ssh commands. Use this to see what a task would do do without running any of the system commands.
This option runs the tasks like the live runner does and disables all system commands like the --dry-run
option does. However, this option also disables all logging and tracks all tasks and sub-tasks that are run. It then outputs the tree of tasks that was run:
TODO: show task tree output example
Use this to show the user all the tasks/sub-tasks that are run and which parent tasks are running them.
This option runs tasks showing more verbose output/details (ie it sets the logger's stdout log level to 'debug'). All Task log_debug
messages will be shown on stdout with this option. This option also makes the stdout and file logs identical.
Dk stores settings in a config. To modify the settings add a configure block:
require 'dk'
Dk.configure do
# settings go here...
end
There are a number of DSL settings that can be configured. Use these to set param values, set default ssh settings, configure tasks that can be called from the CLI, etc.
require 'dk'
Dk.configure do
set_param 'app_name', 'myapp'
set_param 'number_of_things', 5
# ...
end
Use the set_param
method to set new global param values. Any tasks that are run will have these param values available to them using the tasks's params
helper method.
You can configure tasks as callbacks on other tasks using these helper methods:
require 'dk'
Dk.configure do
before MyTask, MyBeforeTask
prepend_before MyTask, MyOtherBeforeTask, 'some_param' => 'some_value'
after MyTask, MyAfterTask, 'some_param' => 'some_value'
prepend_after MyTask, MyOtherAfterTask
end
Each callback can be optionally configured with a set of params. Callbacks can either be appended or prepended (this can be especially useful when you want to control the order 3rd-party tasks are run in).
The callback tasks will be run in the order they are added before/after the run!
method of the task they are added to. The halt
task helper does not stop these callbacks from running.
require 'dk'
Dk.configure do
ssh_hosts 'all_servers', '1.example.com',
'2.example.com',
'3.example.com',
'[email protected]',
'[email protected]'
ssh_hosts 'primary_server', '1.example.com'
ssh_hosts 'web_servers', '1.example.com',
'2.example.com'
ssh_hosts 'db_server', '3.example.com'
ssh_hosts 'bg_servers', '[email protected]',
'[email protected]'
# these are custom args to use on all SSH cmds
ssh_args "-o ForwardAgent=yes "\
"-o ControlMaster=auto "\
"-o ControlPersist=60s "\
"-o UserKnownHostsFile=/dev/null "\
"-o StrictHostKeyChecking=no "\
"-o ConnectTimeout=10 "\
"-o LogLevel=quiet "
# these two hosts use custom SSH ports
host_ssh_args '[email protected]', '-p 12345'
host_ssh_args '[email protected]', '-p 12345'
end
Use the ssh_hosts
method to define a named set of hosts. These hosts can now be referred to by name when running ssh
cmds in tasks.
Use the ssh_args
and host_ssh_args
methods to configure ssh cmd arguments to apply to all SSH cmds. The host-specific args are only applied to SSH cmds run on the specific host.
require 'dk'
Dk.configure do
log_pattern '[%-5l] : %m\n' # [INFO] : blah blah\n"
log_file "log/dk.log" # log all task run details to this file
log_file_pattern '[%d %-5l] : %m\n' # [<datetime> INFO] : blah blah\n"
end
Use the log_pattern
and log_file_pattern
methods to override the default log entry format for stdout and file logging respectively. By default, stdout logs just log the message while the file logs log the date level and message. Uese these to tweak as desired. Dk uses Logsly for its loggers so check out its README for details on using patterns.
Use the log_file
method to turn on file logging. If given a file path, Dk will log all task run details (verbosely) to the file. This is handy when problems occur running tasks in non-verbose mode. File logging is always done verbosely so you can use this to refere to the details of previous task runs even though you may not have seen these details in your stdout logs.
require 'dk'
Dk.configure do
task 'deploy', MyDeployTask
task 'deploy:setup' MyDeploySetupTask
# ...
end
Use the task
method to configure tasks that are runnable via the CLI. This also will display information about the task when using the `--help option in the CLI.
Note: only tasks configured using this method will be runnable from the CLI.
require 'dk'
Dk.configure do
stub_dry_tree_cmd("ls -la") do |spy| # spy is a Local::CmdSpy
spy.stdout = "..."
end
stub_dry_tree_ssh("ls -la") do |spy| # spy is a Remote::CmdSpy
spy.exitstatus = 1 # simulare the ssh call failing
end
# ...
end
Use the stub_dry_tree_cmd
and stub_dry_tree_ssh
methods to add cmd and ssh stubs when using --dry-run
or --tree
. This can be used to control the stdout, stderr or exitstatus of a command. This is useful when a task uses the output of one command for another command or when the task does different logic depending on if a command succeeds or fails.
The Dk::Task
mixin provides a bunch of helper methods to make writing tasks easier and for commonizing common functions.
require 'dk/task'
class MyTask
include Dk::Task
def run!
param?('some_param') #=> true
params['some_param'] #=> "some value"
try_param('some_param') #=> "some value"
param?('some_other_param') #=> false
params['some_other_param'] #=> Dk::NoParamError
try_param('some_other_param') #=> nil
set_param('some_other_param', 'some other value')
param?('some_other_param') #=> true
params['some_other_param'] #=> "some other value"
try_param('some_other_param') #=> "some other value"
end
end
Use the param
helper to access named params that the task was run with. Params can be added two ways: globally from the main config or from callback definitions and run_task
calls.
Use the set_param
method to set new global param values like you would on the main config. Any subsequent tasks that are run will have these param values available to them.
Use the param?
method to check whether a specific param has been set. Prefer this over params.key?
as the task params have special logic for interacting with any runner params that makes the traditional key?
method unreliable.
By default, the params will raise an exception if you try and access a missing key (it will raise Dk::NoParamerror). The idea is that Dk will loudly notify you if you are accessing a param you expect to be set and it is not set. However, this may not be appropriate for every scenario, therefore you can use the try_param
helper and it will just return nil
if the param is not set.
Like in the Config, you can add tasks as callbacks on other tasks using these helper methods:
require 'dk/task'
class MyTask
include Dk::Task
def run!
before SomeTask, MyBeforeTask
prepend_before SomeTask, MyOtherBeforeTask, 'some_param' => 'some_value'
after SomeTask, MyAfterTask, 'some_param' => 'some_value'
prepend_after SomeTask, MyOtherAfterTask
end
end
Each callback can be optionally configured with a set of params. Callbacks can either be appended or prepended (this can be especially useful when you want to control the order 3rd-party tasks are run in). Once a callback is added, it works just as it would if added via the Config.
Use the ssh_hosts
helper to set new ssh host lists values like you would on the main config:
require 'dk/task'
class MyTask
include Dk::Task
def run!
ssh_hosts('my-app-servers') # => nil
ssh_hosts('my-web-servers') # => nil
ssh_hosts 'my-app-servers', '1.example.com'
ssh_hosts 'my-web-servers', '2.example.com', '3.example.com'
ssh_hosts('my-app-servers') # => ['1.example.com']
ssh_hosts('my-web-servers') # => ['2.example.com', '3.example.com']
end
end
Any subsequent tasks that are run will have these ssh hosts available to their ssh
commands.
Use the run_task
helper to run other tasks:
require 'dk/task'
require 'my_other_task'
class MyTask
include Dk::Task
def run!
# ... do something before ...
run_task MyOtherTask, 'some_param' => 'some value'
# ... do something after ...
end
end
This method takes an optional set of param values. Any params given will be merged onto the global config params and made available to just the task being run.
Use the cmd
and cmd!
helpers to run local system cmds:
require 'dk/task'
class MyTask
include Dk::Task
def run!
cmd "ls -la", :env => { 'PWD' => '/path/to/some/dir' }
cmd! "test -d some_file"
end
end
Pass them a string system command to run and it runs it using Scmd. You can optionally pass in an :env
param with any ENV vars that need to be set. A Dk::Local::Cmd
object is returned so you can access data such as the stdout
, stderr
and whether the command was successful or not.
The cmd!
helper is identical to the cmd
helper except that it raises a Dk::Task::CmdRunError
if the command was not successful.
Use the start
helper to asynchronously run a local system cmd:
require 'dk/task'
class MyTask
include Dk::Task
def run!
# run, but don't block waiting on an exitstatus
server_cmd = start "bin/server"
server_cmd.running? # => true
server_cmd.pid # => 12345
server_cmd.exitstatus # => nil
# do other stuff...
server_cmd.wait # wait indefinitely until cmd exits
server_cmd.running? # => false
server_cmd.pid # => 12345
server_cmd.exitstatus # => 0
end
end
OR, you can also asynchorously run with timeouts:
# run, but don't block waiting on an exitstatus
server_cmd = start "bin/server"
begin
server_cmd.wait(10)
rescue Dk::CmdTimeoutError => err
cmd.stop(10) # attempt to stop the cmd nicely, kill if doesn't stop in time
cmd.kill # OR, just kill the cmd now
end
This helper behaves just like cmd
and cmd!
. The only difference is it won't block waiting for the system command to run. It uses Scmd to run system commmands and proxies most of its command API - see the Scmd usage docs for additional reference.
require 'dk/task'
class MyTask
include Dk::Task
def run!
hosts = ['1.example.com', '[email protected]']
ssh "ls -la", :hosts => hosts
ssh! "test -d some_file", :hosts => hosts
end
end
Use the ssh
helper to run remote system cmds on hosts using ssh. Like the normal cmd*
helpers, pass it a string system command and it runs the command on each host using by creating a local system cmd that runs ssh. Like cmd*
you can optionally pass in an :env
param with any ENV vars that need to be set locally. Similare to cmd*
, a Dk::Remote::Cmd
object is returned so you can access whether the command was successful or not.
The ssh!
helper is identical to the ssh
helper except that it raises a Dk::Task::SSHRunError
if the command was not successful.
require 'dk/task'
class MyTask
include Dk::Task
def run!
cmd! "test -d some_file", :dry_tree_run => true
ssh! "test -d some_file", :dry_tree_run => true
end
end
Use this option to force the cmd/ssh to always run, even using the --dry-run
and --tree
CLI options (which put Scmd in test mode and don't run any cmds by default). The idea is that some cmds are essential to the running of the task and if they don't actually run, the task will error out. This is handy for tasks that don't change anything they just read data and that data is used to determine how a task should proceed. This allows you to still get the advantages of the dry run and tree CLI options.
require 'dk/task'
class MyTask
include Dk::Task
def run!
# ... do something ...
halt if nothing_more_to_do
# ... otherwise keep going ...
end
end
Use the halt
helper to stop executing the current task. This doesn't halt the entire run (callbacks, subsequent tasks, etc). It only halts the current task. If you want to stop running everything, just raise an exception.
require 'dk/task'
class MyTask
include Dk::Task
def run!
log_info "will always show up in the CLI output"
log_debug "will only show up in the CLI output in verbose mode"
log_error "will always show, but has special error handling"
end
end
Use the log_*
helpers to log information as the task is running. Each corresponds to a logger level. The CLI logs to stdout on the INFO level by default. When run in verbose mode, it logs to stdout on the DEBUG level. If an optional log file has been configured, the CLI will log to the file on DEBUG level regardless of any verbose mode setting.
You can optionally style your log messages:
log_info "my message", :red, :on_white
log_info "my " + Dk::Ansi.styled_msg("special", :bold, :blue) + " message"
See Dk::Ansi::CODES.keys
for a list of available style names.
require 'dk/task'
class MyTask
include Dk::Task
desc "This is a task that does something"
def run!
# ... do something ...
end
end
The descriptions of any routed tasks will be displayed when running --help
in the CLI.
You can configure other tasks as callbacks to your task:
require 'dk/task'
class MyTask
include Dk::Task
before MyBeforeTask
prepend_before MyOtherBeforeTask, 'some_param' => 'some_value'
after MyAfterTask, 'some_param' => 'some_value'
prepend_after MyOtherAfterTask
def run!
# ... do something ...
end
end
Each callback can be optionally configured with a set of params. These tasks will be run in the order they are added before/after the run!
method of the current task. The halt
helper does not stop these callbacks from running.
You can configure a default list of hosts to use for ssh commands made in a Task. These hosts will be used on all commands that don't specify a custom :hosts
option:
require 'dk/task'
class MyTask
include Dk::Task
ssh_hosts ['1.example.com', '[email protected]']
def run!
ssh('ls -la') # => will ssh to 1.example.com and [email protected]
ssh('ls -la', :hosts => ['3.example.com']) # => will ssh to 3.example.com
end
end
You can also specify a named list of hosts that have been configured:
require 'dk/task'
class MyTask
include Dk::Task
ssh_hosts 'my-app-servers'
def run!
ssh('ls -la') # => will ssh to the configured 'my-app-servers' hosts
end
end
You can also specify a proc that will be instance eval'd on the task instance:
require 'dk/task'
class MyTask
include Dk::Task
ssh_hosts{ params['some-servers'] }
def run!
ssh('ls -la') # => will ssh to the hosts specified by the 'some-servers' param
end
end
require 'dk/task'
class MyTask
include Dk::Task
run_only_once true
def run!
# ... do something ...
end
end
Use this setting to ensure that a task only runs once no matter how many other tasks or sub-tasks either run it manually or configure it as a callback. This is useful for "gatekeeper" tasks such as tasks that validate params, etc. By default, tasks can run multiple times.
Dk comes with a test runner and some test helpers to assist in unit testing tasks. The test runner doesn't run any callback tasks and captures spies for the task, cmd and ssh calls it runs. It also turns off any logging.
# in your test file or whatever
include Dk::Task::TestHelpers
test "my task should do something" do
runner = test_runner(MyTask)
runner.run
runner.runs #=> [TaskRun, Local::CmdSpy, Remote::CmdSpy, ... ]
# make assertions that the logic you expect to run actually ran
task = runner.task #=> MyTask instance
# make assertions about your task instance if needed
end
Dk's test helpers include methods for stubbing cmd and ssh calls. This allows controlling cmd/ssh behavior such as exit status, stdout and stderr. If a task expects to use the stdout/stderr or does special handling when a call errors then the stubbing methods can be used for testing.
# in your test file or whatever
include Dk::Task::TestHelpers
test "my task should do something" do
runner = test_runner(MyTask)
runner.stub_cmd("ls -la") do |spy| # spy is a Local::CmdSpy
spy.stdout = "..."
end
runner.stub_ssh("ls -la", :input => 'whatever') do |spy| # spy is a Remote::CmdSpy
spy.exitstatus = 1 # simulare the ssh call failing
end
# ....
end
You can even stub with procs. Those procs will be instance eval'd against the task to determine if the stub is a match. This allows you to stub the same way the call is made in the task (which is handy when the cmds are driven by dynamic values on the task instance).
# in your test file or whatever
include Dk::Task::TestHelpers
test "my task should do something" do
runner = test_runner(MyTask)
runner.stub_cmd(proc{ self.class.some_cmd_str }, :opts => proc{ some_opts }) do |spy| # spy is a Local::CmdSpy
spy.stdout = "..."
end
runner.stub_ssh("ls -la", :input => proc{ some_task_input_val }) do |spy| # spy is a Remote::CmdSpy
spy.exitstatus = 1 # simulare the ssh call failing
end
# ....
end
These gems extend Dk for specific behavior:
- dk-pkg: Dk logic for installing pkgs
- dk-abdeploy: Dk tasks that implement the A/B deploy scheme
- dk-dumpdb: Build Dk tasks to dump and restore your databases
(if you build your own gem that extends Dk, let us know and we'll link it here)
Add this line to your application's Gemfile:
gem 'dk'
And then execute:
$ bundle
Or install it yourself as:
$ gem install dk
- Fork it
- Create your feature branch (
git checkout -b my-new-feature
) - Commit your changes (
git commit -am 'Added some feature'
) - Push to the branch (
git push origin my-new-feature
) - Create new Pull Request