Vaquero -- What is it?

# Vaquero

What is it?

- Initial discussions between Cisco Cloud team and Comcast in December 2015.

- Scope included container orchestration, OpenStack-based virtualization, system provisioning.

- POC project kicked off in early 2016 under moniker _Gemini_.

- _Nascent effort started by handful of Cisco engineers._

- _Intended as unified deployment system for clustered software._

- _Primary use case: Ansible as CM tool to install clustered software (Kubernetes) on prepared O/S._

.center[
![:scale 30%](https://www.ansible.com/hubfs/2016_Images/Assets/SVG/Ansible-Mark-RGB_Outline-Black.svg?t=1479260166498) ![:scale 30%](https://cdn.rawgit.com/kubernetes/kubernetes/master/logo/logo.svg)]

]

???

Gemini project focused on enhancements to Ansible playbooks used to deploy
Kubernetes on a variety of platforms. Work in [kubernetes/contrib](https://github.com/kubernetes/contrib)

Artifacts: https://github.com/gemini-project/

Required: Prepared O/S = installed, networked, user accounts provisioned, etc.

---

.left-column[
  ## Where did it start?
  ## Where did it go?
]
.right-column[
- Follow-up meetings with Comcast **V.blue[IP]ER** team in early 2016.

- Discussed perceived weaknesses in current workflow with existing tools.

- Effort refocused around provisioning of bare metal and immutable BM infrastructure in May 2016.
]

---

.left-column[
  ## Where did it start?
  ## Where did it go?
]
.right-column[
- Follow-up meetings with Comcast **V.blue[IP]ER** team in early 2016.

- Discussed perceived weaknesses in current workflow with existing tools.

- Effort refocused around provisioning of bare metal and immutable BM infrastructure in May 2016.

Projected adopted name **Vaquero**.red[\*] to reflect new focus.

# Requirements

---

## High-level Requirements

1. Declarative textual representation of bare metal compute.

2. Lowest level of software layer (O/S, Kubernetes, etc.) should be immutable.

3. Prefer Git as repository for description of cluster.

4. System maintenance (upgrades, fault recovery, etc.) will be performed by node reboot.
  - Workflow derived from high-level policies.

## Non-requirements

- Semantic understanding of software layer.
- Dynamic system inventory.

.footnote[More detailed documentation [available](https://ciscocloud.github.io/vaquero-docs/docs/current/requirements.html).]

???

This approach allows the system to determine best steps for rendering
configuration onto compute.

Reboot policies encompass several concepts:

1. reboot one node of this type at a time in a cluster
  2. this node type A depends on node type B completed
  3. allows injection of customized logic via containers

---

## Development and Operational Goals

Vaquero tooling should be easy to use.

- Functionally complete in single binary, and packaged in a Docker container

- Narrow set of external dependencies -- etcd, GitHub

- Familiar command-line interface, taking cues from CLIs such as Git, kubectl, docker, etc.

- Built on powerful foundation of open source tools -- Go, Google's [netboot](https://github.com/google/netboot) package, Docker, etcd, etc.

.pull-right[
  ![:scale 100%](https://cdn.rawgit.com/coreos/etcd/master/logos/etcd-horizontal-color.svg)] .pull-left[
  ![:scale 100%](img/GitHub_Logo.svg)]
---

# Data Model

---

---

**`coreos-pxe.yml`**
```yaml
id: coreos-1053.2.0-stable
name: CoreOS Stable 1053.2.0
major_version: '1053'
minor_version: '2.0'
os_family: CoreOS
release_name: stable
boot:
  kernel: "{{.env.agentURL}}/files/{{.boot.os.release_name}}/{{.boot.os.major_version}}/{{.boot.os.minor_version}}/coreos_production_pxe.vmlinuz"
  initrd:
  - "{{.env.agentURL}}/files/coreos_production_pxe_image.cpio.gz"
cmdline:
  coreos.autologin: ''
```
]

---

**`boot/etcd-master.yml`**
```yaml
id: etcd-master
name: Etcd Master

operating_system: coreos-1053.2.0-stable

unattended:
  type: ignition
  use: etcd.yml

metadata:
  fleet_role: etcd
```

**`boot/clevos.yml`**
```yaml
id: clevos
name: ClevOS 3.7

operating_system: clevos-basic

unattended:
  type: untyped
  use: clevos-answers.provision

metadata:
  root_pass: ********
```
]

---

.left-column[
## Operating system
## Netboot
## Environment
]
.right-column[
**`site/detriot/env.yml`**
```yaml
name: Virtualenv Dev
id: test-site
subnets:
  - id: default
    cidr: 10.10.10.0/24
    domain: foo.bar.com
    gateway: 10.10.10.1
    dns:
    - 8.8.8.8
    ntp:
    - 10.10.10.1
  - id: mgmt-subnet
    cidr: 10.9.9.0/24
    dns:
    - 8.8.8.8
    ntp:
    - 10.10.10.1
metadata:
  name: detroit-preprod
  kind: pre-prod  
  etcd_initial_cluster: node1=http://10.10.10.10:2380,node2=http://10.10.10.11:2380,node3=http://10.10.10.12:2380
  ssh_authorized_keys: []
```
]

---

.left-column[
## Operating system
## Netboot
## Environment
## Inventory
]
.right-column[
**`site/detriot/inventory.yml`**
```yaml
name: host1
*interfaces:
  - type: physical
    subnet: default
    mac: 00:00:00:00:00:01
    bmc:
        type: ipmi
        username: adam
        password: ********
    identifier: eth0
    ipv4: 10.10.10.10
metadata:
    etcd_name: node1
workflow: etcd-master
---
name: host2
interfaces:
  - type: physical
    subnet: default
    mac: 00:00:00:00:00:02
    identifier: eth0
    ipv4: 10.10.10.11
metadata:
    etcd_name: node2
workflow: etcd-master
```
]
---
layout: false
class: center, middle, inverse

# Architecture & Design

---

## Software Subsystems

.center[![:scale 90%](https://cdn.rawgit.com/CiscoCloud/vaquero-docs/master/docs/current/nov16Arch.png)]

???

---

## High-Availability

.center[![:scale 90%](https://cdn.rawgit.com/CiscoCloud/vaquero-docs/master/docs/current/nov16HA.png)]

???