Repository files navigation

Fabric is a command-line tool for application deploymentand systems administration tasks. It allows the user to run commands across aserver farm.

This is intended to be setup and run on your local workstation/laptop.

To install the dependencies:

$ pip install -Ur requirements.txt

NB: if you get a "pip: command not found" error, run this first:

$ sudo easy_install pip

Configure it (seethe fabric documentation for more examples),

$ echo 'user = jimbob' >> ~/.fabricrc

You can view a list of the available tasks:

$ fab -l

And execute against an environment and set of hosts like so:

$ fab integration all hosts...$ fab integration class:frontend do:'uname -a'$ fab integration class:cache,bouncer do:uptime...

Fabric tasks can be run on groups of machines in a variety of different ways.

by puppet class

# target all machines that have the 'govuk::safe_to_reboot::yes' class$ fab integration puppet_class:govuk::safe_to_reboot::yes do:'uname -a'

by numeric machine suffix

# target all machines that end in '2'$ fab integration numbered:2 do:'uname -a'

by node type (as defined in puppet)

# target all 'frontend' machines$ fab integration node_type:frontend do:'uname -a'

by the node name

# target just one node$ fab production -H backend-3.backend do:'uname -a'

It was previously possible to use these scripts from the jumpboxes via thegovuk_fab wrapper. This is no longer supported as it relied on SSH agentforwarding, which is a Bad Thing because other people with root access tothat jumpbox would be able to re-use any of the private keys in your agent.

You should convert to the [local usage](#Local usage) method described aboveand ensure that you do not have the following option anywhere in your~/.ssh/config:

ForwardAgent yes

An example

fab <env> -H '<src_db>' postgresql.sync:<db_name>,<dst_db> -A

the -A must be specified to forward the agent

This will sync the specified database<db_name> from the machine with thehostname of<src_db> to the machine with hostaname<dst_db>. It will destroydata on the destination db