Table of Contents generated with DocToc
- Overview
- TL;DR
- Use case and example
- Technical Details
- Creating a task manifest file
- Add the hosts block
- Add the vars block
- Populate the vars block - defaults
- Populate the vars block - define global options
- Populate the vars block - define subcommands
- Populate the vars block - help/message
- Populate the vars block - embedded shell functions
- Populate the vars block - dynamic inventory expression
- Populate the vars block - inventory file
- Add tasks
- Usage Examples
- Installation
- Appendix
- TODO - Add more tests!
- License and Credits
ansible-taskrunner is a cli app that is essentially an ansible wrapper.
It reads an ansible playbook file as its input, which serves as a task manifest.
If no task manifest is specified, the app will search for 'Taskfile.yaml' in the current working directory.
The inspiration for the tool comes from the gnu make command, which operates in similar fashion, i.e.
- A Makefile defines available build steps
- The make command consumes the Makefile at runtime and exposes these steps as command-line options
If you are running this tool from Windows, please read the section on Bastion Mode
- Ever wanted to add custom switches to the
ansible-playbookcommand? Something like this:
ansible-playbook -i myinventory.txt -d dbhost1 -w webhost1 -t value1 myplaybook.yaml - Well, you can through the use of an ansible-playbook wrapper
- That's where
taskscomes in:
tasks -s -b bar -f foo1
translates to:
ansible-playbook -i /tmp/ansible-inventory16xdkrjd.tmp.ini \ -e "{'some_foo_variable':'foo1'}" -e "{'some_bar_variable':'bar'}" -e "{'playbook_targets':'local'}" Taskfile.yaml
- Jump down to the usage examples to see this in action
- Review the installation instructions if you want to test-drive it
- Read on if you want to dig deeper into the tool
- An enterprise-grade application named contoso-app
- Multiple teams:
- Development
- Engineering
- DBA
- Operations
- QA
- Ansible is the primary means of invoking business and operational processes across the numerous environments
You must ensure all teams adopt a standardized approach to running ansible workloads
Upon investigating the current approach, you observe the following:
- Users tend to create wrapper scripts that call the ansible-playbook command
- These scripts don't follow any naming convention, as you've noted:
- run.sh
- start.sh
- playbook.sh
- These shell scripts have common attributes:
- Dynamically populate ansible-playbook variables via the --extra-vars option
- Dynamically creating ansible inventories
- Performing pre/post-flight tasks
- Providing a command-line interface
Advantages to the above approach:
- Quick-n-dirty, anyone can get started relatively quickly with writing ansible automation
Disadvantages:
- Lack of standards:
- Leads to difficulty in collaboration and code refactoring
- Decreased re-usability of codebase
- This design encourages standalone playbooks
- Makes it more difficult to package actions as roles
- Duplicate efforts across codebase
Employ a pre-execution script that operates at a layer above the ansible-playbook command:
- Accomplishes the same as the above, but in more uniform manner
- Support for custom command-line parameters/flags
- Dynamic inventory expression
- Embedded shell functions
Advantages to this approach:
- Easier to manage
- If you know YAML and Ansible, you can get started relatively quickly with writing ansible automation
- Single executable (/usr/local/bin/tasks)
Disadvantages:
- Target ansible controller needs to have the
taskscommand installed
As stated in the overview, this tool functions much like the make command in that it accepts an input file that essentially extends its cli options.
We create a specially formatted ansible-playbook that serves as a task manifest file (by default, Taskfile.yaml).
This task manifest file:
- Extends the
taskscommand - Is a valid ansible playbook (Taskfile.yaml), and can thus be launched with the
ansible-playbookcommand - Variables available to the pre-execution phase are also available to the ansible execution phase
In the following sections, we'll be building a sample manifest/playbook.
Start by opening up your favorite text/IDE/editor and creating a new task manifest file named Taskfile.yaml.
Add hosts, gather_facts, etc:
Click to Expand
Taskfile.yaml
### The hosts block
- hosts: myhosts
gather_facts: true
become: true
Remember, the task runner will ultimately be calling the ansible-playbook command against
this very same file,
so it must be a valid ansible playbook.
Let's add the 'vars' block, which allows us to populate some default values:
Click to Expand
Taskfile.yaml
### The hosts block
- hosts: myhosts
gather_facts: true
become: true
### The vars block
vars:
Let's add some default variables to the playbook:
Click to Expand
Taskfile.yaml
### The hosts block
- hosts: myhosts
gather_facts: true
become: true
### The vars block
vars:
var1: value1
var2: value2
var3: value3
var4: |-
This is a multi-line value
of type string
var5:
- listvalue1
- listvalue2
- listvalue3
- listvalue4
var6:
some_key:
some_child_key: dictvalue1
var7: $(echo some_value)
var8: 8000
dbhosts:
- dbhost1
- dbhost2
- dbhost3
webhosts:
- webhost1
- webhost2
- webhost3
As you can see, we've defined a number of variables holding different values.
The rules for evaluation of these are as follows:
Variable | Ansible Evaluation | Shell Function Evaluation
-------------------------------------------- | ----------------------- | -----------------------
str_var: value1 | String | String
bool_var: True | Boolean | String
num_var: 3 | Integer | Integer
multiline_var: | | Multiline String | String (heredoc)
This is a multi-line value | |
of type string | |
list_var: | List Object | String (heredoc)
- item1 | |
- item2 | |
dict_var: | Dictionary Object | JSON String (heredoc)
key1: somevalue1 | |
key2: somevalue2 | |
shell_var: $(grep somestring /some/file.txt) | String | Depends on output
Additionally, arguments supplied from the command-line conform
to the type specified in the options definition, with "string" being the default type.
Global options are available to all subcommands.
These are defined under the vars.globals.options key.
Let's add a simple example:
Click to Expand
Taskfile.yaml
### The hosts block
- hosts: myhosts
gather_facts: true
become: true
### The vars block
vars:
var1: value1
var2: value2
var3: value3
var4: |-
This is a multi-line value
of type string
var5:
- listvalue1
- listvalue2
- listvalue3
- listvalue4
var6:
some_key:
some_child_key: dictvalue1
var7: $(echo some_value)
var8: 8000
dbhosts:
- dbhost1
- dbhost2
- dbhost3
webhosts:
- webhost1
- webhost2
- webhost3
globals:
options:
my_global_option:
help: "This is my global option"
short: -g
long: --global
var: some_global_variable
Next, we define the available subcommands and their options.
Let's add a sub-command named run along with its command-line options:
Click to Expand
Taskfile.yaml
### The hosts block
- hosts: myhosts
gather_facts: true
become: true
### The vars block
vars:
var1: value1
var2: value2
var3: value3
var4: |-
This is a multi-line value
of type string
var5:
- listvalue1
- listvalue2
- listvalue3
- listvalue4
var6:
some_key:
some_child_key: dictvalue1
var7: $(echo some_value)
var8: 8000
dbhosts:
- dbhost1
- dbhost2
- dbhost3
webhosts:
- webhost1
- webhost2
- webhost3
### Global Options Block
globals:
options:
my_global_option:
help: "This is my global option"
short: -g
long: --global
var: some_global_variable
### The commands block
commands:
run:
options:
foo:
help: "This is some foo option"
short: -f
long: --foo
type: choice
var: some_foo_variable
required: True
not_required_if:
- some_bar_variable
options:
- foo1
- foo2
bar:
help: "This is some bar option"
short: -b
long: --bar
type: str
var: some_bar_variable
required: False
required_if:
- hello
- some_baz_variable
baz:
help: "This is some baz option"
short: -z
long: --baz
type: str
var: some_baz_variable
required: False
mutually_exclusive_with:
- some_bar_variable
- some_foo_variable
envvar:
help: "The value for this argument can be derived from an Environmental Variable"
short: -E
long: --env-var
type: str
var: env_var
env_var: SOME_ENVIRONMENT_VARIABLE
env_var_show: True
num:
help: "This is a numeric argument"
short: -n
long: --number
var: some_num_variable
type: int
required: False
env_var_show: True
targets:
help: "Playbook targets"
short: -t
long: --targets
type: str
var: playbook_targets
required: True
multiple:
help: |-
This option can be specified multiple times
short: -m
long: --multiple
type: str
var: multiple_arg
allow_multiple: True
some_switch:
help: |-
This is some boolean option, behaves like Click's switches,
holds the value of True if specified
see: https://github.com/pallets/click
short: -s
long: --some-switch
is_flag: true
var: some_switch
required: True
say_hello:
help: "Invoke the 'hello' embedded shell function"
short: -hello
long: --say-hello
type: str
var: hello
is_flag: True
say_goodbye:
help: "Invoke the 'goodbye' embedded shell function"
short: -goodbye
long: --say-goodbye
type: str
var: goodbye
is_flag: True
hidden_option:
help: "This is a hidden option"
short: -O
long: --hidden-option
is_hidden: True
type: str
var: hidden
is_flag: True
verbose:
help: |-
This is a sample paramter that supports counting, as with:
-v, -vv, -vvv, which would evaluate to 1, 2, and 3, respectively
short: -v
allow_counting: True
var: verbosity
As you can see, commands are defined via YAML, and the syntax is mostly self-explanatory.
Currently, the parameters available to any given option are
consistent with click version 8.1.x, see API — Click Documentation (8.1.x)
Important Notes:
- An option's var key:
- In the case of standard options, this variable holds the value of the arguments passed to the option
- In the case of flags/switches, this variable is a Boolean
- The variable is available during the entire runtime
- In the above example, the
-helloand-goodbyeoptions point to special mapped
variables that themselves map to corresponding shell functions defined in the subcommand's
functions directive. We'll discuss this more in section embedded-shell-functions.
As I mentioned before, the above mapped variables can be used during runtime.
That is, they can be referenced in any defined shell functions,
dynamic inventory expression logic, as well as during ansible execution.
Consider the -f|-foo from the example.
Whatever argument you pass to this option becomes the value for the mapped variable some_foo_variable.
Again, this variable is made available to the underlying subprocess call, and thus to ansible.
So when we call the tasks command like so tasks run -f foo1, the value for the some_foo_variable becomes foo.
Next, we add the help/message section:
Click to Expand
Taskfile.yaml
### The hosts block
- hosts: myhosts
gather_facts: true
become: true
### The vars block
vars:
var1: value1
var2: value2
var3: value3
var4: |-
This is a multi-line value
of type string
var5:
- listvalue1
- listvalue2
- listvalue3
- listvalue4
var6:
some_key:
some_child_key: dictvalue1
var7: $(echo some_value)
var8: 8000
dbhosts:
- dbhost1
- dbhost2
- dbhost3
webhosts:
- webhost1
- webhost2
- webhost3
### Global Options Block
globals:
options:
my_global_option:
help: "This is my global option"
short: -g
long: --global
var: some_global_variable
### The commands block
commands:
run:
options:
foo:
help: "This is some foo option"
short: -f
long: --foo
type: choice
var: some_foo_variable
required: True
not_required_if:
- some_bar_variable
options:
- foo1
- foo2
bar:
help: "This is some bar option"
short: -b
long: --bar
type: str
var: some_bar_variable
required: False
required_if:
- hello
- some_baz_variable
baz:
help: "This is some baz option"
short: -z
long: --baz
type: str
var: some_baz_variable
required: False
mutually_exclusive_with:
- some_bar_variable
- some_foo_variable
envvar:
help: "The value for this argument can be derived from an Environmental Variable"
short: -E
long: --env-var
type: str
var: env_var
env_var: SOME_ENVIRONMENT_VARIABLE
env_var_show: True
num:
help: "This is a numeric argument"
short: -n
long: --number
var: some_num_variable
type: int
required: False
env_var_show: True
targets:
help: "Playbook targets"
short: -t
long: --targets
type: str
var: playbook_targets
required: True
multiple:
help: |-
This option can be specified multiple times
short: -m
long: --multiple
type: str
var: multiple_arg
allow_multiple: True
some_switch:
help: |-
This is some boolean option, behaves like Click's switches,
holds the value of True if specified
see: https://github.com/pallets/click
short: -s
long: --some-switch
is_flag: true
var: some_switch
required: True
say_hello:
help: "Invoke the 'hello' embedded shell function"
short: -hello
long: --say-hello
type: str
var: hello
is_flag: True
say_goodbye:
help: "Invoke the 'goodbye' embedded shell function"
short: -goodbye
long: --say-goodbye
type: str
var: goodbye
is_flag: True
hidden_option:
help: "This is a hidden option"
short: -O
long: --hidden-option
is_hidden: True
type: str
var: hidden
is_flag: True
verbose:
help: |-
This is a sample paramter that supports counting, as with:
-v, -vv, -vvv, which would evaluate to 1, 2, and 3, respectively
short: -v
allow_counting: True
var: verbosity
help:
message: |
Invoke the 'run' command
epilog: |
This line will be displayed at the end of the help text message
examples:
- example1: |
tasks $command
- example2: |
Usage example 2
Running tasks run --help should return the list of parameters along with the help message you defined.
Add embedded shell functions:
Taskfile.yaml
### The hosts block
- hosts: myhosts
gather_facts: true
become: true
### The vars block
vars:
var1: value1
var2: value2
var3: value3
var4: |-
This is a multi-line value
of type string
var5:
- listvalue1
- listvalue2
- listvalue3
- listvalue4
var6:
some_key:
some_child_key: dictvalue1
var7: $(echo some_value)
var8: 8000
dbhosts:
- dbhost1
- dbhost2
- dbhost3
webhosts:
- webhost1
- webhost2
- webhost3
### Global Options Block
globals:
options:
my_global_option:
help: "This is my global option"
short: -g
long: --global
var: some_global_variable
### The commands block
commands:
run:
options:
foo:
help: "This is some foo option"
short: -f
long: --foo
type: choice
var: some_foo_variable
required: True
not_required_if:
- some_bar_variable
options:
- foo1
- foo2
bar:
help: "This is some bar option"
short: -b
long: --bar
type: str
var: some_bar_variable
required: False
required_if:
- hello
- some_baz_variable
baz:
help: "This is some baz option"
short: -z
long: --baz
type: str
var: some_baz_variable
required: False
mutually_exclusive_with:
- some_bar_variable
- some_foo_variable
envvar:
help: "The value for this argument can be derived from an Environmental Variable"
short: -E
long: --env-var
type: str
var: env_var
env_var: SOME_ENVIRONMENT_VARIABLE
env_var_show: True
num:
help: "This is a numeric argument"
short: -n
long: --number
var: some_num_variable
type: int
required: False
env_var_show: True
targets:
help: "Playbook targets"
short: -t
long: --targets
type: str
var: playbook_targets
required: True
multiple:
help: |-
This option can be specified multiple times
short: -m
long: --multiple
type: str
var: multiple_arg
allow_multiple: True
some_switch:
help: |-
This is some boolean option, behaves like Click's switches,
holds the value of True if specified
see: https://github.com/pallets/click
short: -s
long: --some-switch
is_flag: true
var: some_switch
required: True
say_hello:
help: "Invoke the 'hello' embedded shell function"
short: -hello
long: --say-hello
type: str
var: hello
is_flag: True
say_goodbye:
help: "Invoke the 'goodbye' embedded shell function"
short: -goodbye
long: --say-goodbye
type: str
var: goodbye
is_flag: True
hidden_option:
help: "This is a hidden option"
short: -O
long: --hidden-option
is_hidden: True
type: str
var: hidden
is_flag: True
verbose:
help: |-
This is a sample paramter that supports counting, as with:
-v, -vv, -vvv, which would evaluate to 1, 2, and 3, respectively
short: -v
allow_counting: True
var: verbosity
### The help message
help:
message: |
Invoke the 'run' command
epilog: |
This line will be displayed at the end of the help text message
examples:
- example1: |
tasks $command
- example2: |
Usage example 2
### Embedded shell functions
functions:
hello:
shell: bash
help: Say hello
source: |-
echo Hello! The value for var1 is $var1
goodbye:
shell: bash
help: Say goodbye
source: |-
echo The value for var1 is $var1. Goodbye!
Again, notice the two switches -hello and -goodbye, with mapped variables hello and goodbye, respectively.
These mapped variables correspond to keys in the functions block with matching names.
As such, specifying either or both -hello and -goodbye in your tasks run invocation
will short-circuit normal operation and execute the corresponding functions
in the order in which you call them.
Try it yourself by running:
tasks run -t local -s -b bar -m one -m two -vvv -O -hello -goodbyetasks run -t local -s -b bar -m one -m two -vvv -O -goodbye -hello
There is also a special flag ---invoke-function that is globally available to all subcommands.
Invocation is as follows: tasks <subcommand> ---invoke-function <function_name>.
In our example, we would run: tasks run -t local -s -b bar -m one -m two -vvv -O ---invoke-function hello
For more usage examples, see the appendix.
Let's briefly side-step into embedded shell functions.
The syntax for nesting these under the functions key is as follows:
name_of_function:
shell: bash, ruby, or python
help: Help Text to Display
hidden: false/true
source: |-
{{ code }}
hello:
shell: bash
help: Hello World in Bash
hidden: false
source: |-
echo 'Hello World!'
hello:
shell: python
help: Hello World in Python
hidden: false
source: |-
print('Hello World!')
hello:
shell: ruby
help: Hello World in Ruby
hidden: false
source: |-
puts 'Hello World!'
A useful feature of this tool is the ability to define your ansible
inventory as a dynamic expression in the Taskfile itself.
To do so, we populate the with the inventory_expression key.
When the inventory is defined in this manner, the logic is as follows:
- The inventory expression is evaluated
- An ephemeral inventory file is created with the
contents of this file being the output, or result, of that expression - The fully qualified path to the ephemeral inventory file is specified as the
argument to theansible-playbookinventory parameter-i
Let's define our inventory expression:
Click to Expand
Taskfile.yaml
### The hosts block
- hosts: myhosts
gather_facts: true
become: true
### The vars block
vars:
var1: value1
var2: value2
var3: value3
var4: |-
This is a multi-line value
of type string
var5:
- listvalue1
- listvalue2
- listvalue3
- listvalue4
var6:
some_key:
some_child_key: dictvalue1
var7: $(echo some_value)
var8: 8000
dbhosts:
- dbhost1
- dbhost2
- dbhost3
webhosts:
- webhost1
- webhost2
- webhost3
### Global Options Block
globals:
options:
my_global_option:
help: "This is my global option"
short: -g
long: --global
var: some_global_variable
### The commands block
commands:
run:
options:
foo:
help: "This is some foo option"
short: -f
long: --foo
type: choice
var: some_foo_variable
required: True
not_required_if:
- some_bar_variable
options:
- foo1
- foo2
bar:
help: "This is some bar option"
short: -b
long: --bar
type: str
var: some_bar_variable
required: False
required_if:
- hello
- some_baz_variable
baz:
help: "This is some baz option"
short: -z
long: --baz
type: str
var: some_baz_variable
required: False
mutually_exclusive_with:
- some_bar_variable
- some_foo_variable
envvar:
help: "The value for this argument can be derived from an Environmental Variable"
short: -E
long: --env-var
type: str
var: env_var
env_var: SOME_ENVIRONMENT_VARIABLE
env_var_show: True
num:
help: "This is a numeric argument"
short: -n
long: --number
var: some_num_variable
type: int
required: False
env_var_show: True
targets:
help: "Playbook targets"
short: -t
long: --targets
type: str
var: playbook_targets
required: True
multiple:
help: |-
This option can be specified multiple times
short: -m
long: --multiple
type: str
var: multiple_arg
allow_multiple: True
some_switch:
help: |-
This is some boolean option, behaves like Click's switches,
holds the value of True if specified
see: https://github.com/pallets/click
short: -s
long: --some-switch
is_flag: true
var: some_switch
required: True
say_hello:
help: "Invoke the 'hello' embedded shell function"
short: -hello
long: --say-hello
type: str
var: hello
is_flag: True
say_goodbye:
help: "Invoke the 'goodbye' embedded shell function"
short: -goodbye
long: --say-goodbye
type: str
var: goodbye
is_flag: True
hidden_option:
help: "This is a hidden option"
short: -O
long: --hidden-option
is_hidden: True
type: str
var: hidden
is_flag: True
verbose:
help: |-
This is a sample paramter that supports counting, as with:
-v, -vv, -vvv, which would evaluate to 1, 2, and 3, respectively
short: -v
allow_counting: True
var: verbosity
### The help message
help:
message: |
Invoke the 'run' command
epilog: |
This line will be displayed at the end of the help text message
examples:
- example1: |
tasks $command
- example2: |
Usage example 2
### Embedded shell functions
functions:
hello:
shell: bash
help: Say hello
source: |-
echo Hello! The value for var1 is $var1
goodbye:
shell: bash
help: Say goodbye
source: |-
echo The value for var1 is $var1. Goodbye!
### The inventory expression
inventory_expression: |
[local]
localhost ansible_connection=local
[web_hosts]
$(echo -e "${webhosts}" | tr ',' '\n')
[db_hosts]
$(echo -e "${dbhosts}" | tr ',' '\n')
[myhosts:children]
web_hosts
db_hosts
As you can see, the inventory expression is dynamic, as
it gets evaluated based on the output of inline shell commands.
Let's focus on the variable $webhosts.
As per the logic described above, the variable $webhosts is a heredoc with a value of:
webhosts='
webhost1
webhost2
webhost3
'
As such, the web_hosts group in the inventory expression ...
[web_hosts]
$(echo -e "${webhosts}" | tr ',' '\n')
... will evaluate to:
[web_hosts]
webhost1
webhost2
webhost3
Also, notice how the inline shell command transforms commas into newline characters by way of the transform (tr) command.
This makes it so that if we were to have defined the webhosts variable
in the Tasksfile as webhosts: webhost1,webhost2,webhost3, it would have had the same outcome
as defining it as a list object in the vars block.
We can specify an inventory file instead of an inventory expression with the inventory_file key.
Click to Expand
Taskfile.yaml
### The hosts block
- hosts: myhosts
gather_facts: true
become: true
### The vars block
vars:
var1: value1
var2: value2
var3: value3
var4: |-
This is a multi-line value
of type string
var5:
- listvalue1
- listvalue2
- listvalue3
- listvalue4
var6:
some_key:
some_child_key: dictvalue1
var7: $(echo some_value)
var8: 8000
dbhosts:
- dbhost1
- dbhost2
- dbhost3
webhosts:
- webhost1
- webhost2
- webhost3
### Global Options Block
globals:
options:
my_global_option:
help: "This is my global option"
short: -g
long: --global
var: some_global_variable
### The commands block
commands:
run:
options:
foo:
help: "This is some foo option"
short: -f
long: --foo
type: choice
var: some_foo_variable
required: True
not_required_if:
- some_bar_variable
options:
- foo1
- foo2
bar:
help: "This is some bar option"
short: -b
long: --bar
type: str
var: some_bar_variable
required: False
required_if:
- hello
- some_baz_variable
baz:
help: "This is some baz option"
short: -z
long: --baz
type: str
var: some_baz_variable
required: False
mutually_exclusive_with:
- some_bar_variable
- some_foo_variable
envvar:
help: "The value for this argument can be derived from an Environmental Variable"
short: -E
long: --env-var
type: str
var: env_var
env_var: SOME_ENVIRONMENT_VARIABLE
env_var_show: True
num:
help: "This is a numeric argument"
short: -n
long: --number
var: some_num_variable
type: int
required: False
env_var_show: True
targets:
help: "Playbook targets"
short: -t
long: --targets
type: str
var: playbook_targets
required: True
multiple:
help: |-
This option can be specified multiple times
short: -m
long: --multiple
type: str
var: multiple_arg
allow_multiple: True
some_switch:
help: |-
This is some boolean option, behaves like Click's switches,
holds the value of True if specified
see: https://github.com/pallets/click
short: -s
long: --some-switch
is_flag: true
var: some_switch
required: True
say_hello:
help: "Invoke the 'hello' embedded shell function"
short: -hello
long: --say-hello
type: str
var: hello
is_flag: True
say_goodbye:
help: "Invoke the 'goodbye' embedded shell function"
short: -goodbye
long: --say-goodbye
type: str
var: goodbye
is_flag: True
hidden_option:
help: "This is a hidden option"
short: -O
long: --hidden-option
is_hidden: True
type: str
var: hidden
is_flag: True
verbose:
help: |-
This is a sample paramter that supports counting, as with:
-v, -vv, -vvv, which would evaluate to 1, 2, and 3, respectively
short: -v
allow_counting: True
var: verbosity
### The help message
help:
message: |
Invoke the 'run' command
epilog: |
This line will be displayed at the end of the help text message
examples:
- example1: |
tasks $command
- example2: |
Usage example 2
### Embedded shell functions
functions:
hello:
shell: bash
help: Say hello
source: |-
echo Hello! The value for var1 is $var1
goodbye:
shell: bash
help: Say goodbye
source: |-
echo The value for var1 is $var1. Goodbye!
### Inventory file
inventory_file: '/some/path/some/inventory.yaml'
Notes of Importance:
- The value you provide to the inventory_file key supports templating
- That is, any of the variables available runtime variables can be used, for example:
inventory_file: '/some/path/some/inventory_$foo_variable.yaml'inventory_file: '/some/path/some/inventory_$var1.yaml'
- That is, any of the variables available runtime variables can be used, for example:
- You should not be specifying both an inventory_file and an inventory_expression, as you get unexpected results.
Finally, let's add some proper ansible tasks!
Click to Expand
Taskfile.yaml
### The hosts block
- hosts: myhosts
gather_facts: true
become: true
### The vars block
vars:
var1: value1
var2: value2
var3: value3
var4: |-
This is a multi-line value
of type string
var5:
- listvalue1
- listvalue2
- listvalue3
- listvalue4
var6:
some_key:
some_child_key: dictvalue1
var7: $(echo some_value)
var8: 8000
dbhosts:
- dbhost1
- dbhost2
- dbhost3
webhosts:
- webhost1
- webhost2
- webhost3
### Global Options Block
globals:
options:
my_global_option:
help: "This is my global option"
short: -g
long: --global
var: some_global_variable
### The commands block
commands:
run:
options:
foo:
help: "This is some foo option"
short: -f
long: --foo
type: choice
var: some_foo_variable
required: True
not_required_if:
- some_bar_variable
options:
- foo1
- foo2
bar:
help: "This is some bar option"
short: -b
long: --bar
type: str
var: some_bar_variable
required: False
required_if:
- hello
- some_baz_variable
baz:
help: "This is some baz option"
short: -z
long: --baz
type: str
var: some_baz_variable
required: False
mutually_exclusive_with:
- some_bar_variable
- some_foo_variable
envvar:
help: "The value for this argument can be derived from an Environmental Variable"
short: -E
long: --env-var
type: str
var: env_var
env_var: SOME_ENVIRONMENT_VARIABLE
env_var_show: True
num:
help: "This is a numeric argument"
short: -n
long: --number
var: some_num_variable
type: int
required: False
env_var_show: True
targets:
help: "Playbook targets"
short: -t
long: --targets
type: str
var: playbook_targets
required: True
multiple:
help: |-
This option can be specified multiple times
short: -m
long: --multiple
type: str
var: multiple_arg
allow_multiple: True
some_switch:
help: |-
This is some boolean option, behaves like Click's switches,
holds the value of True if specified
see: https://github.com/pallets/click
short: -s
long: --some-switch
is_flag: true
var: some_switch
required: True
say_hello:
help: "Invoke the 'hello' embedded shell function"
short: -hello
long: --say-hello
type: str
var: hello
is_flag: True
say_goodbye:
help: "Invoke the 'goodbye' embedded shell function"
short: -goodbye
long: --say-goodbye
type: str
var: goodbye
is_flag: True
hidden_option:
help: "This is a hidden option"
short: -O
long: --hidden-option
is_hidden: True
type: str
var: hidden
is_flag: True
verbose:
help: |-
This is a sample paramter that supports counting, as with:
-v, -vv, -vvv, which would evaluate to 1, 2, and 3, respectively
short: -v
allow_counting: True
var: verbosity
### The help message
help:
message: |
Invoke the 'run' command
epilog: |
This line will be displayed at the end of the help text message
examples:
- example1: |
tasks $command
- example2: |
Usage example 2
### Embedded shell functions
functions:
hello:
shell: bash
help: Say hello
source: |-
echo Hello! The value for var1 is $var1
goodbye:
shell: bash
help: Say goodbye
source: |-
echo The value for var1 is $var1. Goodbye!
### The inventory expression
inventory_expression: |
[local]
localhost ansible_connection=local
[web_hosts]
$(echo -e "${webhosts}" | tr ',' '\n')
[db_hosts]
$(echo -e "${dbhosts}" | tr ',' '\n')
[myhosts:children]
web_hosts
db_hosts
tasks:
- name: Show Variables
debug:
msg: |-
{{ hostvars[inventory_hostname] | to_nice_json }}
The task above will display all available host variables.
Quick usage examples:
- Display help for main command
tasks --help - Display help for the run subcommand
tasks run --help - Initialize your workspace
tasks init - Run the Taskfile.yaml playbook, passing in additional options to the underlying subprocess
tasks run -t local -s -b bar -m one -m two - Don't do anything, just echo the underlying shell command
tasks run -t local -s -b bar -m one -m two -O ---echo
Result should be similar to:
ansible-playbook -i /var/folders/5f/4g4xnnv958q52108qxd2rj_r0000gn/T/ansible-inventorytlmz2hpz.tmp.ini \ -e "{'var1':'${var1}'}" ... Taskfile.yaml - Run the embedded function
hello
tasks run -t local -s -b bar -m one -m two -hello - Run the embedded functions
helloandgoodbye
run -t local -s -b bar -m one -m two -hello -goodbye
Ansible-taskrunner consists of the tasks command.
It can be installed in a few ways:
- pip install ansible-taskrunner
- pip install git+https://github.com/berttejeda/ansible-taskrunner.git
- Obtaining a release (these lag behind the pip distributions)
Note: You'll need to pre-install a python distribution for the Windows MSI release. Not yet sure if I am doing something wrong or if that's by design. I lean toward the former :|
Review the examples directory for more hands-on usage samples.
When you pass the --- options separator to any subcommand, anything
after the separator is passed directly to the ansible subprocess.
If you're launching the tasks command from a Windows host, this tool will automatically execute in Bastion Mode
Under Bastion Mode, the tasks command will:
- Execute the
ansible-playbooksubprocess via a bastion host, i.e. a remote machine that should haveansibleinstalled - This is done via ssh using the paramiko module
Running in Bastion Mode requires a configuration file containing the ssh connection settings.
To initialize this configuration file, you can simply run tasks init.
For full usage options, enter in tasks init --help.
Once you've initialized the configuration file, you should see sftp-config.json in your workspace.
This configuration file is fashioned after the sftp
plugin for Sublime Text and is thus compatible.
If you define the playbook variable ansible_playbook_command, this will override the underlying ansible-playbook command invocation.
As an example, suppose I define this variable in the above Taskfile.yaml, as follows:
- hosts: myhosts
gather_facts: true
become: true
vars:
ansible_playbook_command: 'python ${HOME}/ansible_2.7.8/ansible-playbook'
var1: value1
var2: value2
var3: value3
# ...
Upon invoking the tasks command with the ---echo flag:
- The temporary inventory would be revealed as:
inventory_is_ephemeral=True
if [[ "$inventory_is_ephemeral" == "True" ]];then
echo -e """${inventory_expression}"""| while read line;do
eval "echo -e ${line}" >> "/var/folders/some/path/ansible-inventoryo4fw4ttc.tmp.ini";
done
fi;
*The above inventory file path will differ, of course
- And the underlying shell command would be revealed as:
python ${HOME}/ansible_2.7.8/ansible-playbook \
-i /var/folders/some/path/ansible-inventoryo4fw4ttc.tmp.ini \
-e "{'var1':'${var1}'}" \
-e "{'var2':'${var2}'}" \
-e "{'var3':'${var3}'}" \
... Taskfile.yaml
Anything defined under the pre_execution variable will be evaluated before all other statements in the underlying shell expression.
As an example, suppose I define the pre_execution variable in the above Taskfile.yaml, as follows:
- hosts: myhosts
gather_facts: true
become: true
vars:
pre_execution: |-
export pxe_var=some_value
touch /tmp/.run.lock
Upon invoking the tasks command with the ---echo flag:
- The underlying shell expression would be revealed as:
...
export pxe_var=some_value
touch /tmp/.run.lock
...
The commands above are always placed before
all variables declarations in the underlying shell expresison.
This is similar to pre_execution, except that anything defined under the post_execution variable will be evaluated after all other statements in the underlying shell expression.
By defining the playbook dictionary variable environment_vars,
the following occurs:
- For each dictionary
key: valuepair:- A corresponding
exportstatement is defined in the underlying shell expression
- A corresponding
As an example, suppose I define this variable in the above Taskfile.yaml, as follows:
- hosts: myhosts
gather_facts: true
become: true
vars:
ansible_playbook_command: 'python ${HOME}/ansible_2.7.8/ansible-playbook'
var1: value1
var2: value2
var3: value3
some_path: /some/path
environment_vars:
MY_ENV_VAR1: "${some_path}/${var1}"
MY_ENV_VAR2: "${some_path}/${var2}"
# ...
Upon invoking the tasks command with the ---echo flag:
- The underlying shell expression would be revealed as:
var1="value1"
var2="value2"
export MY_ENV_VAR1="${some_path}/${var1}"
export MY_ENV_VAR2="${some_path}/${var2}"
These export statements are always placed after
all variables declarations in the underlying shell expresison.
Any variables matching ANSIBLE_.* will automatically be expressed as export statements.
As an example, suppose I define such variables in the above Taskfile.yaml, as follows:
- hosts: myhosts
gather_facts: true
become: true
vars:
var1: value1
var2: value2
ANSIBLE_VAULT_PASSWORD_FILE: /some/path/some/password_file
ANSIBLE_CALLBACK_PLUGINS: /some/other/path/some/plugins
# ...
Upon invoking the tasks command with the ---echo flag:
- The underlying shell expression would be revealed as:
var1="value1"
var2="value2"
export ANSIBLE_VAULT_PASSWORD_FILE="/some/path/some/password_file"
export ANSIBLE_CALLBACK_PLUGINS="/some/other/path/some/plugins"
You can override the underlying command-line provider in two ways:
- Via the tasks config file (see examples)
- By defining the variable cli_provider in the specified Taskfile
As an example, suppose I define this variable in the above Taskfile.yaml, as follows:
- hosts: myhosts
gather_facts: true
become: true
vars:
cli_provider: bash
# ...
Upon invoking the tasks command, you will note that the app no longer operates in an ansible-playbook mode, but rather as yaml-abstracted bash-script.
There are three cli-providers built in to the tasks command:
- ansible
- bash
- vagrant
Apart from utilizing the --- options separator, you can specify additional options to pass to the underlying ansible-playbook subprocess by setting an appropriate value for __ansible_run_flags__
via Environmental variable or Taskfile, as with:
__ansible_run_flags__: --diff
or
export __ansible_run_flags__='--diff'
The __tasks_file__ variable points to the current Taskfile.
It is available to the underlying subprocess shell.
The __command__ variable points to the name of the invoked subcommand.
It is available to the underlying subprocess shell.
This tool supports the following advanced options:
- Mutually Exclusive, see Mutually exclusive option groups in python Click - Stack Overflow.
- Mutually Inclusive
- Conditionally required
Suppose you want a set of options such that:
- You want to accept one option but only if another, related option has not been specified
You can accomplish this by defining your option with the following parameters:
- required: False
- mutually_exclusive_with:
- some_bar_variable
- some_foo_variable
In the above configuration, calling this option along with options
-f|-foo and -b|-bar will trigger an illegal usage error, since you've
marked the option as mutually exclusive with either of these two options.
Feel free to review the Taskfile.yaml, as you'll find an example of:
- mutually exclusive
- mutually inclusive
- conditionally required
There might exist edge cases in which you need two different subcommands to be defined with identical commandline options.
This could be useful for, say, subcommands like install or uninstall,
where the only differentiating factor between the two would be the
name of the subcommand being invoked.
To enable this, simply declare your subcommand with a pipe | delimiter, as with:
commands:
install|uninstall:
options:
foo:
help: "Install/Uninstall an app"
short: -n
long: --app-name
type: choice
var: app_name
required: True
As of version 1.1.5, simple templating is available to the following objects:
- Help messages
- Examples
- Options
- Options values
What this means is that we expose a limited set of internal variables to the above.
As an example:
help:
message: |
Invoke the 'run' command
epilog: |
This line will be displayed at the end of the help text message
examples:
- example1: |
tasks -f $tf_path --foo foo --bar bar
- example2: |
tasks -f $tf_path --foo foo --baz baz
In the above strings, $tf_path will expand to the internal variable tf_path,
which holds the relative path to the current tasks file.
Below is a list of available variables for your convenience:
Variable | Description
------------- | -------------
__command__ | The name of the current subcommand
cli_args | The current command-line invocation
cli_args_short | The current command-line invocation, minus the executable
sys_platform | The OS Platform as detected by Python
tf_path | The relative path to the specified Taskfile
Additionally, all currently set environmental variables are available for templating.
This script also ships as a zipapp executable (similar to a windows .exe).
Head over to the releases page for release downloads.
You can also build your own single-executable zipapp, as follows:
- Make sure you have the make-zipapp executable in your path
- Invoking build tasks
- Build zipapp:
python ansible_taskrunner/cli.py -f Makefile.yaml run ---make zipapp
Read More on zipapps: zipapp — Manage executable Python zip archives — Python 3.7.4rc2 documentation
To run all tests, simply call the test script, as with:
python tests/test_ansible_taskrunner.py
This project adopts the the MIT distribution License.
Releases come bundled with the following opensource python packages:
Lastly, this package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.
- Cookiecutter: https://github.com/audreyr/cookiecutter
- audreyr/cookiecutter-pypackage: https://github.com/audreyr/cookiecutter-pypackage