An asynchronous deployment and operations platform with the aim to better enable simplicity, faster time to delivery, and consistency.
This project is maintained by oshied
Orchestration files allow directord to run any number of jobs against a given set of targets. The specification file is simple and made to be easy. Each file is loaded as an array and can contain many jobs.
The values available within an orchestration file are targets and jobs.
async Optional is a boolean. Enabling this option allows all tasks
within a given orchestration to run asynchronously.Asynchronous orchestrations allow operators to run many orchestrations at the same time while following the defined task ordering.
jobs is an array of hashes.
name is a String. An orchestration can be named. This is done through
the use of the name key. When orchestrations are named, both the job
list and fingerprint output will use the defined name in the returned
information.
targets is an array of strings.
assign is an array which allows operators to define given set of
targets. This is useful when defining node assignments outside of
the typical targets consruct.This is the basic, syntax.
---
- targets: []
jobs: []
async: False
Within orchestration file the “targets” key is optional. If this key is undefined, Directord will run against all available targets.
---
- jobs: []
Running jobs requires each item be a key=value pair. The key is a VERB used for the given command and the value is the execution.
---
- jobs:
- RUN: "echo hello world"
Each job can use either inline or YAML arguments. In the following example the orchestration is using a YAML argument to instruct the command to skip the cache.
---
- jobs:
- RUN: "echo hello world"
vars:
skip_cache: true
Several CLI options are available when orchestrating a deployment, such as
--restrictand--ignore-cache. These option provide for the ability to replay specific tasks or forcefully ignore the client side cache.
This example orchestration file will copy the local client ssh keys from the
directory /home/centos/.ssh to all nodes within the clister. Then, on the
three noted targets, wget will be installed.
---
- jobs:
- WORKDIR: /home/centos/.ssh
- RUN: chmod 600 /home/centos/.ssh/* && chmod 700 /home/centos/.ssh
- ADD: /home/centos/.ssh/* /home/centos/.ssh/
vars:
chown: "centos:centos"
- RUN: --skip-cache echo hello world
- targets:
- df.next-c1.localdomain-client-1
- df.next-c2.localdomain-client-1
- df.next-c3.localdomain-client-1
jobs:
- RUN: dnf install -y wget
Upon the execution of each job, a new UUID will be presented to you for tracking purposes. This information can be used with the manage commands.
Running an orchestration is simple and follows the same pattern as running an adhoc execution.
$ directord orchestrate ${ORCHESTRATION_FILE_NAME} [<switches> ...]
Each job within the jobs list is a dictionary item with the fist key
ALWAYS representing the verb used within the given job. This verb corresponds
to a known component.
All jobs can make use extra options to enhance the user experience. This options are useful when defining complex variables, which may be difficult to express inline or within a “string” format.
assign is a list object which allows operators to override the
given set of targets for a particular job.
name is a string object which provides a human friendly name
for a job definition. This name can be used for an improved UX when
looking into job debug information.
vars is a dictionary of arguments that will be passed back into the
execution options for a given job