Defining Layouts

Learn the layout specification and how to create your own provisioning layouts.

Create a file ubuntu.py:

"""layout spec"""

from __future__ import annotations

from ogc.deployer import init
from ogc.fs import expand_path
from ogc.log import get_logger

log = get_logger("ogc")

deployment = init(
    layout_model=dict(
        instance_size="e2-standard-4",
        name="ubuntu-ogc",
        provider="google",
        remote_path="/home/ubuntu/ogc",
        runs_on="ubuntu-2004-lts",
        scale=9,
        scripts="fixtures/ex_deploy_ubuntu",
        username="ubuntu",
        ssh_private_key=expand_path("~/.ssh/id_rsa_libcloud"),
        ssh_public_key=expand_path("~/.ssh/id_rsa_libcloud.pub"),
        ports=["22:22", "80:80", "443:443", "5601:5601"],
        tags=[],
        labels=dict(
            division="engineering", org="obs", team="observability", project="perf"
        ),
    ),
)

Each layout has a friendly name associated as seen by ubuntu-ogc. The next section is going to go over each option and describe its meaning.

provider

Define which cloud the layout will operate in. Currently supported options are aws and google.

runs-on

Define the base OS image to be deployed on to the nodes. The current supported list of names are:

AWS	Google
ubuntu-2004	ubuntu-2004
ubuntu-1804	ubuntu-1804
ubuntu-1604	ubuntu-1604
sles-15	sles-15
sles-12	sles-12
sles-11	sles-11
debian-10	debian-10
debian-9	debian-9
debian-8	debian-8
	rhel-8
	rhel-7
	rhel-6

instance-size

Define the machine size, this is dependent on which provider is chosen. The instance-size correlates with the instance size naming for each cloud.

For example, on AWS you would use instance-size = "c5.4xlarge" and in Google's case, instance-size = "e2-standard-4".

username

The ssh user to use when deploying and accessing the nodes. This is also somewhat dependent on which provider is used.

In the case of Google, any username can be given. In the case of AWS, the base machines have a pre-loaded user that must be used:

AWS	Username
centos	centos
debian	admin
oracle	ec2-user
sles	ec2-user
ubuntu	ubuntu
windows1	ogc

Caution

A lot of cloud machine images disable root login, try to avoid using that as a user and utilize sudo for anything requiring elevated permissions.

scripts

The location on your machine where templates/scripts resides. These will be uploaded and executed during the deployment phase.

Note

See scripting for more information.

scale

How many nodes of each layout to deploy. This is also referenced during a deployment reconciliation phase.

remote-path (optional)

If set, any uploads/downloads outside of what's defined in scripts will be placed in that remote path.

include (optional)

A list of files/wildcards to include in the upload

exclude (optional)

A list of files/wildcards to exclude in the upload

artifacts (optional)

The remote path where script execution output is stored. This is used when pulling artifacts with ogc pull-artifacts and also utilized during node teardown. This will download any artifacts found into artifacts/instance-name/.

tags (optional)

Define tags for each layout, allows additional filtering capabilities and deployment options when used with ogc ls and ogc exec

ports (optional)

Define what ingress ports are available when accessing the node.

---

1. This is the default user for our contributed packer build for Windows ↩