Airsloop is a two node site deployment reference and continuous integration pipeline. The goal of this site is to be used as a reference for mini Airship deployments with one controller and one or more compute nodes.
Airsloop is meant to get an operator up and running quickly with an Airship lab environment, and has some resiliency and security features tuned down. Please see the Seaworthy manifests for a production-oriented example.
The site manifests are available at site/airsloop.
Airsloop pipeline automates deployment flow documented in Site Authoring and Deployment Guide.
The pipeline is implemented as Jenkins Pipeline (Groovy), see code for the pipeline at Jenkinsfile.
The manifest software version overrides (versions.yaml) are setup to deploy OpenStack Ocata.
The versions are kept up to date via updater.py, a utility that updates versions.yaml latest charts and (selected) images.
The pipeline attempts to uplift and deploy latest versions on daily basis.
While HW configuration is flexible, Airsloop reference manifests reflect a single controller and a single compute node. The aim of this is to create a minimalistic lab/demo reference environment.
Increasing the number of compute nodes will require site overrides to align parts of the system such as Ceph osds, etcd, etc.
This site is stripped down from all the complicated hardware and configuration requirements that Airship Seaworthy site is using and that leads to simplified deployments from disk,networking and hardware perspective.
Airsloop site has 2 bare-metal servers: 1 controller, and 1 compute node. See host profiles for the servers here.
- Control (masters)
- Compute (workers)
For simplification of the configuration, we recommend using hostnames provided below.
For a two node deployment the nodes and their hostnames are:
Physical (underlay) networks are described in Drydock site configuration here. It defines OOB (iLO/IPMI), untagged PXE, and multiple tagged general use networks. Also no bond interfaces are used in Airsloop deployment.
In this deployment the networking is simplified compared to Airship Seaworthy site. There are only two NICs required (excluding oob), one for PXE and another one for the rest of the networks separated using VLAN segmentation. The recommendation is to use the highest bandwidth device available to carry to carry the data networks.
Below is an example of network configuration:
Calico overlay for k8s POD networking uses IPIP mesh.
Because Airsloop is a minimalistic deployment the required number of disks is just one per node. That disk is not only used by the OS but also by Ceph Journals and OSDs. The way that this is achieved is by using directories and not extra disks for Ceph storage. Ceph OSD configuration can be changed in a Ceph chart override.
Host profiles in Airship are tightly coupled with the hardware profiles. That means every disk or interface which is described in host profiles should have a corresponding reference to the hardware profile which is being used.
Airship always identifies every NIC or disk by its PCI or SCSI address and that means that the interfaces and the disks that are defined in host and hardware profiles should have the correct PCI and SCSI addresses objectively.
Let’s give an example by following the host profile of Airsloop site.
In this Host Profile is defined that the slave interface that will be used for the pxe boot will be the pxe_nic01. That means a corresponding entry should exist in this Hardware Profile which it does. So when drydock and maas try to deploy the node it will identify the interface by the PCI address that is written in the Hardware profile.
A simple way to find out which PCI od SCSi address corresponds to which NIC or Disk is to use the lshw command. More inforamtion about that command can be found Here.
This section describes what changes need to be made to the existing manifests of Airsloop for the addition of an extra compute node to the cluster.
First and foremost the user should go to the nodes.yaml file and add an extra section for the new compute node.
The next step is to add a similar section as the existing airsloop-compute-1 section to the pki-catalog.yaml. This is essential for the correct generation of certificates and the correct communication between the nodes in the cluster.
Also every time the user adds an extra compute node to the cluster then the number of OSDs that are managed by this manifest Ceph-client should be increased by one.
Last step is to regenerate the certificates which correspond to this certificates.yaml file so the changes in the pki-catalog.yaml file takes place. This can be done through the promenade CLI.
- TODO: Specify which node(s) the command(s) in this section are run on.
- Also if there is an assumption that we have a node with Ubuntu already provisioned, that assumption or steps should be specified along with any Ubuntu version requirements/assumptions and other pre-requisite steps (e.g., installing NTP)
Below are the steps that a user should follow to deploy the Airsloop site:
- TODO: Add the prerequisite steps that the user needs to do
- before starting executing the below steps such as: installing git , installing docker, clone sevral repos etc.
tools/airship pegleg site -r /target collect airsloop -s collect
tools/airship promenade generate-certs -o /target/certs /target/collect/*.yaml
Generate genesis.sh scipt
tools/airship promenade build-all -o /target/bundle /target/collect/*.yaml /target/certs/*.yaml
Execute the genesis.sh script
- cd /target/bundle
If the genesis.sh script completed succesfully
Deploy site through shipyard
- tools/airship shipyard create configdocs design –directory=/target/collect
- tools/airship shipyard commit configdocs
- tools/airship shipyard create action deploy_site
Check the actions that are already created
tools/shipyard get actions