The structure and content of fittings file¶
Keywords used by plumbery¶
This is the list of keywords that are known by plumbery, and that can be used in a fittings file. You will find some comprehensive example at the bottom of this page.
| Keyword | Required | Description |
|---|---|---|
| apiHost | yes | A private API endpoint. See Selecting a data centre |
| appliance | yes | Name of the image to deploy, .e.g., ‘Ubuntu’. See Selecting an image to deploy |
| backup | no | Cloud backup configuration. See Configuring cloud backup |
| beachhead | no | Advanced networking setting. See How to connect plumbery to remote nodes? |
| blueprints | yes | A collection of blueprints |
| cloud-config | no | The cloud-config details. See Configure nodes with cloud-config |
| cpu | no | The number of CPU, or the CPU configuration. See Configuring CPU and memory |
| default | no | The class of some fittings. See Using defaults |
| defaults | no | A map of default values. See Using defaults |
| description | no | One line of text with hashtags, e.g., ‘This is #ubuntu #master node’ |
| disks | no | Storage type and size. See Configuring virtual disks |
| domain | yes | See Defining network domains |
| ethernet | yes | See Defining Ethernet networks |
| glue | no | See Connecting nodes to the Internet and to networks |
| information | no | A list of strings explaining what the fittings does |
| links | no | A map, including ‘documentation’ and ‘credit’ as possible values |
| listeners | no | A collection of listener objects, describing load balancers settings |
| locationId | yes | Target data centre, e.g., ‘EU6’. See Selecting a data centre |
| memory | no | Amount of RAM in gigabytes. See Configuring CPU and memory |
| monitoring | no | Monitoring plan, either essentials or advanced. No default |
| nodes | yes | A collection of node objects, describing the servers to be deployed |
| parameters | no | Settings that can be provided externally to plumbery |
| regionId | no | Identify API endpoint, e.g., ‘dd-ap’. See Selecting a data centre |
| running | no | If set to always, then plumbery cannot delete the fittings |
Multiple documents in one fittings file¶
YAML allows for multiple documents to be assembled in one fittings plan. The separation of documents is done with three dashes at the beginning of a line. The first document is reserved for plumbery parameters, default settings, etc. Therefore the description of blueprints starts on the second document:
---
information:
- "NFS client and server at two different data centres"
---
blueprints:
...
Deploying in multiple geographies¶
Since Plumbery processes each document independently, it is really easy to configure a deployment that spans multiple data centres, like in the following example:
---
information:
- "Multi-Geography deployment example"
---
regionId: dd-eu
locationId: EU6
blueprints:
...
---
regionId: dd-na
locationId: NA9
blueprints:
...
Combining private and public clouds in a deployment¶
Private MCPs are set using the apiHost parameter, you must also include the datacenter ID of the cloud as the locationId. You can then include another document(s) with the public cloud fittings:
---
information:
- "Multi-Geography deployment example"
---
apiHost: my-private-cloud.com
locationId: MY1
blueprints:
...
---
regionId: dd-na
locationId: NA9
blueprints:
...
Environment variables¶
Environment variables are those set on the system that is running plumbery. If you mention an environment variable in a fitting plan, it will be transmitted and used on the target nodes. You can think about it as a convenient way to share configuration across nodes created by plumbery. On the other end, you may take the risk to expose sensitive information outside your computer.
For example, let say that you deploy a virtual machine that will talk to the Cisco Spark API. Before this, you registered to Cisco developer eco-system and got a token that has been saved in CISCO_SPARK_TOKEN, on your workstation. The most straightforward approach would be to just copy this to the target virtual machine. If this new machine will run Ubuntu, you could say:
write_files:
- path: /etc/profile.d/cisco_spark.sh
permissions: "0755"
content: |
#!/bin/sh
export CISCO_SPARK_TOKEN="{{ environment.CISCO_SPARK_TOKEN }}"
During the processing of the fitting plan, plumbery will replace the variable with the actual value of the token on your machine. Therefore the configuration file actually transmitted to the target machine could be like the following:
write_files:
- path: /etc/profile.d/cisco_spark.sh
permissions: "0755"
content: |
#!/bin/sh
export CISCO_SPARK_TOKEN="YWM2OEG4OGItNTQ5YS00MDU2LThkNWEtMJNkODk3ZDZLOGQ0OVGlZWU1NmYtZWyY"
When the target virtual machine will boot, the token will be made available in every user sessions.
Please note that this is probably a very lazy approach, that does induce security risks. In the situation described before, you should get a separate token for the new server, instead of sharing a secret from your own machine. You have been warned.
Dynamic variables¶
Dynamic variables reflect values assigned by the cloud platform such as network addresses.
| Variable | Example | Description |
|---|---|---|
| Self-name | {{ node.name }} | Name of the current node, e.g., Server1 |
| Self private address | {{ node.private }} | Private IPv4 address, e.g., 10.11.2.3 |
| Self public address | {{ node.public }} | Public IPv4 address, e.g., 8.9.10.11 – requires the directive ‘internet’ to assign an address |
| Self IPv6 address | {{ node.ipv6 }} | IPv6 address defined for the node |
| Node private address | {{ server1.private }} | Private IPv4 address of server named server1 |
| Node public address | {{ server1.public }} | Public IPv4 address – requires the directive ‘internet’ as well |
| Node IPv6 address | {{ host357.ipv6 }} | IPv6 address defined for the node named host357 |
Using dynamic variables¶
Plumbery will set pre-defined attributes when asked, for example, a private IPv4 address to a node. This is working great, and all you have to do for this is document such attributes in a fittings plan.
However in many situations you will handle information that is either created dynamically, or that is declared outside a fittings plan.
Some examples:
- IPv6 addresses assigned automatically by the platform
- IPv4 addresses selected dynamically from subnets
- random password used for the setup of a MySQL server
- SSH keys to be created for a specific deployment
This is where you can use templating capabilities of plumbery directly in the fittings plan.
To illustrate the case we will consider a deployment with two nodes deployed in different data centres. The nodes have to communicate over the IPv6 back-end infrastructure that connects all data centres deployed by Dimension Data. In other terms, the IPv6 address of node-a has to be given to node-b, and the IPv6 address of node-b has to be given to node-a.
As you can expect, the most straightforward implementation relies on the file /etc/hosts of both nodes. This is the natural place where names and addresses can be mapped. In plumbery, we would start with something like the following:
write_files:
# map IPv6 addresses with names
#
- path: /etc/hosts
content: |
{{ node-a.ipv6 }} node-a
{{ node-b.ipv6 }} node-b
Before the content of /etc/hosts is actually sent to the nodes, plumbery looks for references to dynamic variables, and replaces them with actual values. For example:
write_files:
# map IPv6 addresses with names
#
- path: /etc/hosts
content: |
2001:0db8:85a3:0:0:8a2e:370:7334 node-a
2001:db8:85a3:8d3:1319:8a2e:370:7348 node-b
Complete example¶
---
information:
- "Let's Chat server, self-hosted chat for private teams"
parameters:
locationId:
information:
- "the target data centre for this deployment"
type: locations.list
default: EU6
links:
documentation: https://github.com/DimensionDataCBUSydney/plumbery-contrib/tree/master/fittings/collaboration/letschat
credit: https://mborgerson.com/setting-up-lets-chat-on-ubuntu
defaults:
cloud-config:
ssh_keys:
rsa_private: |
{{ key.rsa_private }}
rsa_public: "{{ key.rsa_public }}"
users:
- default
- name: ubuntu
sudo: 'ALL=(ALL) NOPASSWD:ALL'
ssh-authorized-keys:
- "{{ key.rsa_public }}"
- "{{ local.rsa_public }}"
disable_root: true
ssh_pwauth: false
---
locationId: "{{ parameter.locationId }}"
blueprints:
- letschat:
domain:
name: myDomain
ipv4: 2
ethernet:
name: myNetwork
subnet: 10.0.0.0
nodes:
- letschat01:
description: "#chat server #ubuntu"
information:
- "this is the Let's Chat server for our team"
- "browse http://{{ node.public }}:5000/ to enter conversations"
appliance: 'Ubuntu 14'
cpu: 8
memory: 32
disks:
- 1 50 standard
glue:
- internet 22 5000
monitoring: essentials
cloud-config:
hostname: "{{ node.name }}"
packages:
- nodejs
- npm
- nodejs-legacy
- mongodb
- mongodb-server
- git
runcmd:
- echo "===== Growing LVM with added disk"
- pvcreate /dev/sdb
- vgextend rootvol00 /dev/sdb
- lvextend -l +100%FREE /dev/mapper/rootvol00-rootlvol00
- resize2fs /dev/mapper/rootvol00-rootlvol00
- echo "===== Handling ubuntu identity"
- cp -n /etc/ssh/ssh_host_rsa_key /home/ubuntu/.ssh/id_rsa
- cp -n /etc/ssh/ssh_host_rsa_key.pub /home/ubuntu/.ssh/id_rsa.pub
- chown ubuntu:ubuntu /home/ubuntu/.ssh/*
- echo "===== Installing Let's Chat"
- cd /home/ubuntu
- git clone https://github.com/sdelements/lets-chat.git
- cd lets-chat
- npm install
- cp settings.yml.sample settings.yml
- sed -i "/host:/s/'[^']*'/'{{ node.name }}'/" settings.yml
- echo "===== Starting the server"
- npm start