feat!: template all ansible dir and rename template to bootstrap (#847)

This commit is contained in:
Devin Buhl authored and GitHub committed 2023-07-12 11:44:23 -04:00
1 parent c7f10b0c6d
commit c9d988a66f
177 files changed
+238 -129

No files matched your search

+34
View File
@@ -0,0 +1,34 @@
---
- name: Cluster Installation
hosts: localhost
connection: local
vars_files:
- defaults/main.yaml
- vars/config.yaml
- vars/addons.yaml
tasks:
- name: Get absolute path to this Git repository # noqa: command-instead-of-module
ansible.builtin.command: git rev-parse --show-toplevel
changed_when: false
check_mode: false
register: repository
failed_when: repository.rc != 0
- name: Set facts
ansible.builtin.set_fact:
repository_path: "{{ repository.stdout }}"
- name: Verify configuration
ansible.builtin.include_tasks: tasks/validation/main.yaml
- name: Template Sops configuration
ansible.builtin.include_tasks: tasks/sops/main.yaml
- name: Template Ansible configuration
ansible.builtin.include_tasks: tasks/ansible/main.yaml
- name: Template Kubernetes configuration
ansible.builtin.include_tasks: tasks/kubernetes/main.yaml
- name: Template Kubernetes addon configuration
ansible.builtin.include_tasks: tasks/addons/main.yaml
+2
View File
@@ -0,0 +1,2 @@
---
bootstrap_coredns_addr: "{{ bootstrap_service_cidr | ansible.utils.nthhost(10) }}"
@@ -0,0 +1,63 @@
---
- name: Set addon facts
ansible.builtin.set_fact:
addon_name: csi-driver-nfs
addon_namespace: kube-system
- name: Ensure directories exist for {{ addon_namespace }}/{{ addon_name }}
when: item.state == 'directory'
ansible.builtin.file:
path: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path }}"
state: directory
mode: "0755"
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
- name: Template unencrypted files for {{ addon_namespace }}/{{ addon_name }}
when: item.state == 'file' and 'sops' not in item.path
ansible.builtin.template:
src: "{{ item.src }}"
dest: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path | replace('.j2', '') }}"
mode: "0755"
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
# https://github.com/ansible-collections/community.sops/issues/153
- name: Template encrypted files for {{ addon_namespace }}/{{ addon_name }}
block:
- name: Rename Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml"
dest: "{{ repository_path }}/.sops.yaml.orig"
mode: "0755"
- name: Remove Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml"
state: absent
- name: Template encrypted files
when: item.state == 'file' and 'sops' in item.path
community.sops.sops_encrypt:
path: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path | replace('.j2', '') }}"
encrypted_regex: ^(data|stringData)$
age: ["{{ bootstrap_age_public_key }}"]
content_yaml: "{{ lookup('ansible.builtin.template', item.src) | from_yaml }}"
mode: "0755"
force: true
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
rescue:
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
+63
View File
@@ -0,0 +1,63 @@
---
- name: Set addon facts
ansible.builtin.set_fact:
addon_name: grafana
addon_namespace: monitoring
- name: Ensure directories exist for {{ addon_namespace }}/{{ addon_name }}
when: item.state == 'directory'
ansible.builtin.file:
path: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path }}"
state: directory
mode: "0755"
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
- name: Template unencrypted files for {{ addon_namespace }}/{{ addon_name }}
when: item.state == 'file' and 'sops' not in item.path
ansible.builtin.template:
src: "{{ item.src }}"
dest: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path | replace('.j2', '') }}"
mode: "0755"
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
# https://github.com/ansible-collections/community.sops/issues/153
- name: Template encrypted files for {{ addon_namespace }}/{{ addon_name }}
block:
- name: Rename Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml"
dest: "{{ repository_path }}/.sops.yaml.orig"
mode: "0755"
- name: Remove Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml"
state: absent
- name: Template encrypted files
when: item.state == 'file' and 'sops' in item.path
community.sops.sops_encrypt:
path: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path | replace('.j2', '') }}"
encrypted_regex: ^(data|stringData)$
age: ["{{ bootstrap_age_public_key }}"]
content_yaml: "{{ lookup('ansible.builtin.template', item.src) | from_yaml }}"
mode: "0755"
force: true
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
rescue:
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
+63
View File
@@ -0,0 +1,63 @@
---
- name: Set addon facts
ansible.builtin.set_fact:
addon_name: hajimari
addon_namespace: default
- name: Ensure directories exist for {{ addon_namespace }}/{{ addon_name }}
when: item.state == 'directory'
ansible.builtin.file:
path: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path }}"
state: directory
mode: "0755"
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
- name: Template unencrypted files for {{ addon_namespace }}/{{ addon_name }}
when: item.state == 'file' and 'sops' not in item.path
ansible.builtin.template:
src: "{{ item.src }}"
dest: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path | replace('.j2', '') }}"
mode: "0755"
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
# https://github.com/ansible-collections/community.sops/issues/153
- name: Template encrypted files for {{ addon_namespace }}/{{ addon_name }}
block:
- name: Rename Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml"
dest: "{{ repository_path }}/.sops.yaml.orig"
mode: "0755"
- name: Remove Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml"
state: absent
- name: Template encrypted files
when: item.state == 'file' and 'sops' in item.path
community.sops.sops_encrypt:
path: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path | replace('.j2', '') }}"
encrypted_regex: ^(data|stringData)$
age: ["{{ bootstrap_age_public_key }}"]
content_yaml: "{{ lookup('ansible.builtin.template', item.src) | from_yaml }}"
mode: "0755"
force: true
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
rescue:
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
@@ -0,0 +1,63 @@
---
- name: Set addon facts
ansible.builtin.set_fact:
addon_name: kube-prometheus-stack
addon_namespace: monitoring
- name: Ensure directories exist for {{ addon_namespace }}/{{ addon_name }}
when: item.state == 'directory'
ansible.builtin.file:
path: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path }}"
state: directory
mode: "0755"
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
- name: Template unencrypted files for {{ addon_namespace }}/{{ addon_name }}
when: item.state == 'file' and 'sops' not in item.path
ansible.builtin.template:
src: "{{ item.src }}"
dest: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path | replace('.j2', '') }}"
mode: "0755"
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
# https://github.com/ansible-collections/community.sops/issues/153
- name: Template encrypted files for {{ addon_namespace }}/{{ addon_name }}
block:
- name: Rename Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml"
dest: "{{ repository_path }}/.sops.yaml.orig"
mode: "0755"
- name: Remove Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml"
state: absent
- name: Template encrypted files
when: item.state == 'file' and 'sops' in item.path
community.sops.sops_encrypt:
path: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path | replace('.j2', '') }}"
encrypted_regex: ^(data|stringData)$
age: ["{{ bootstrap_age_public_key }}"]
content_yaml: "{{ lookup('ansible.builtin.template', item.src) | from_yaml }}"
mode: "0755"
force: true
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
rescue:
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
@@ -0,0 +1,63 @@
---
- name: Set addon facts
ansible.builtin.set_fact:
addon_name: kubernetes-dashboard
addon_namespace: monitoring
- name: Ensure directories exist for {{ addon_namespace }}/{{ addon_name }}
when: item.state == 'directory'
ansible.builtin.file:
path: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path }}"
state: directory
mode: "0755"
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
- name: Template unencrypted files for {{ addon_namespace }}/{{ addon_name }}
when: item.state == 'file' and 'sops' not in item.path
ansible.builtin.template:
src: "{{ item.src }}"
dest: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path | replace('.j2', '') }}"
mode: "0755"
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
# https://github.com/ansible-collections/community.sops/issues/153
- name: Template encrypted files for {{ addon_namespace }}/{{ addon_name }}
block:
- name: Rename Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml"
dest: "{{ repository_path }}/.sops.yaml.orig"
mode: "0755"
- name: Remove Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml"
state: absent
- name: Template encrypted files
when: item.state == 'file' and 'sops' in item.path
community.sops.sops_encrypt:
path: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path | replace('.j2', '') }}"
encrypted_regex: ^(data|stringData)$
age: ["{{ bootstrap_age_public_key }}"]
content_yaml: "{{ lookup('ansible.builtin.template', item.src) | from_yaml }}"
mode: "0755"
force: true
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
rescue:
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
+25
View File
@@ -0,0 +1,25 @@
---
- name: Process addon csi-driver-nfs
when: csi_driver_nfs.enabled | default(false)
ansible.builtin.include_tasks: csi_driver_nfs.yaml
- name: Process addon hajimari
when: hajimari.enabled | default(false)
ansible.builtin.include_tasks: hajimari.yaml
- name: Process addon grafana
when: grafana.enabled | default(false)
ansible.builtin.include_tasks: grafana.yaml
- name: Process addon kubernetes-dashboard
when: kubernetes_dashboard.enabled | default(false)
ansible.builtin.include_tasks: kubernetes_dashboard.yaml
- name: Process addon kube-prometheus-stack
when: kube_prometheus_stack.enabled | default(false)
ansible.builtin.include_tasks: kube_prometheus_stack.yaml
- name: Process addon weave-gitops
when: weave_gitops.enabled | default(false)
ansible.builtin.include_tasks: weave_gitops.yaml
+63
View File
@@ -0,0 +1,63 @@
---
- name: Set addon facts
ansible.builtin.set_fact:
addon_name: weave-gitops
addon_namespace: flux-system
- name: Ensure directories exist for {{ addon_namespace }}/{{ addon_name }}
when: item.state == 'directory'
ansible.builtin.file:
path: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path }}"
state: directory
mode: "0755"
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
- name: Template unencrypted files for {{ addon_namespace }}/{{ addon_name }}
when: item.state == 'file' and 'sops' not in item.path
ansible.builtin.template:
src: "{{ item.src }}"
dest: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path | replace('.j2', '') }}"
mode: "0755"
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
# https://github.com/ansible-collections/community.sops/issues/153
- name: Template encrypted files for {{ addon_namespace }}/{{ addon_name }}
block:
- name: Rename Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml"
dest: "{{ repository_path }}/.sops.yaml.orig"
mode: "0755"
- name: Remove Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml"
state: absent
- name: Template encrypted files
when: item.state == 'file' and 'sops' in item.path
community.sops.sops_encrypt:
path: "{{ repository_path }}/kubernetes/apps/{{ addon_namespace }}/{{ addon_name }}/{{ item.path | replace('.j2', '') }}"
encrypted_regex: ^(data|stringData)$
age: ["{{ bootstrap_age_public_key }}"]
content_yaml: "{{ lookup('ansible.builtin.template', item.src) | from_yaml }}"
mode: "0755"
force: true
with_community.general.filetree: ["../templates/addons/{{ addon_name }}/"]
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
rescue:
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
+121
View File
@@ -0,0 +1,121 @@
---
- name: Ensure Kubernetes directories exist
when: item.state == 'directory'
ansible.builtin.file:
path: "{{ repository_path }}/ansible/{{ item.path }}"
state: directory
mode: "0755"
with_community.general.filetree: ["../templates/ansible/"]
- name: Template Ansible unencrypted files
when: item.state == 'file' and 'sops' not in item.path and '.DS_Store' not in item.path
ansible.builtin.template:
src: "{{ item.src }}"
dest: "{{ repository_path }}/ansible/{{ item.path | regex_replace('.j2$', '') }}"
mode: "0755"
with_community.general.filetree: ["../templates/ansible/"]
# https://github.com/ansible-collections/community.sops/issues/153
- name: Template Ansible encrypted files
block:
- name: Rename Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml"
dest: "{{ repository_path }}/.sops.yaml.orig"
mode: "0755"
- name: Remove Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml"
state: absent
- name: Template Ansible encrypted files
when: item.state == 'file' and 'sops' in item.path
community.sops.sops_encrypt:
path: "{{ repository_path }}/ansible/{{ item.path | replace('.j2', '') }}"
encrypted_regex: ^(data|stringData)$
age: ["{{ bootstrap_age_public_key }}"]
content_yaml: "{{ lookup('ansible.builtin.template', item.src) | from_yaml }}"
mode: "0755"
force: true
with_community.general.filetree: ["../templates/ansible/"]
- name: Template encrypted node secrets
community.sops.sops_encrypt:
path: "{{ repository_path }}/ansible/inventory/host_vars/{{ item.name }}.sops.yaml"
age: ["{{ bootstrap_age_public_key }}"]
content_yaml: "{{ lookup('ansible.builtin.template', 'templates/node.sops.yaml.j2', template_vars=dict(password=item.password)) | from_yaml }}"
mode: "0755"
force: true
loop: "{{ bootstrap_nodes.master + bootstrap_nodes.worker | default([]) }}"
loop_control:
label: "{{ item.address }}"
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
rescue:
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
# - name: Template Ansible configuration
# ansible.builtin.template:
# src: "{{ item.src }}"
# dest: "{{ repository_path }}/{{ item.dest }}"
# mode: "0755"
# loop:
# - src: templates/ansible/supplemental.yaml.j2
# dest: ./ansible/inventory/group_vars/kubernetes/supplemental.yaml
# - src: templates/ansible/hosts.yaml.j2
# dest: ./ansible/inventory/hosts.yaml
# # https://github.com/ansible-collections/community.sops/issues/153
# - name: Template Ansible encrypted files
# block:
# - name: Rename Sops configuration file
# ansible.builtin.copy:
# src: "{{ repository_path }}/.sops.yaml"
# dest: "{{ repository_path }}/.sops.yaml.orig"
# mode: "0755"
# - name: Remove Sops configuration file
# ansible.builtin.file:
# path: "{{ repository_path }}/.sops.yaml"
# state: absent
# - name: Template encrypted node secrets
# community.sops.sops_encrypt:
# path: "{{ repository_path }}/ansible/inventory/host_vars/{{ item.name }}.sops.yaml"
# age: ["{{ bootstrap_age_public_key }}"]
# content_yaml: "{{ lookup('ansible.builtin.template', 'templates/ansible/node.sops.yaml.j2', template_vars=dict(password=item.password)) | from_yaml }}"
# mode: "0755"
# force: true
# loop: "{{ bootstrap_nodes.master + bootstrap_nodes.worker | default([]) }}"
# loop_control:
# label: "{{ item.address }}"
# - name: Restore Sops configuration file
# ansible.builtin.copy:
# src: "{{ repository_path }}/.sops.yaml.orig"
# dest: "{{ repository_path }}/.sops.yaml"
# mode: "0755"
# - name: Remove backed up Sops configuration file
# ansible.builtin.file:
# path: "{{ repository_path }}/.sops.yaml.orig"
# state: absent
# rescue:
# - name: Restore Sops configuration file
# ansible.builtin.copy:
# src: "{{ repository_path }}/.sops.yaml.orig"
# dest: "{{ repository_path }}/.sops.yaml"
# mode: "0755"
# - name: Remove backed up Sops configuration file
# ansible.builtin.file:
# path: "{{ repository_path }}/.sops.yaml.orig"
# state: absent
+58
View File
@@ -0,0 +1,58 @@
---
- name: Ensure Kubernetes directories exist
when: item.state == 'directory'
ansible.builtin.file:
path: "{{ repository_path }}/kubernetes/{{ item.path }}"
state: directory
mode: "0755"
with_community.general.filetree: ["../templates/kubernetes/"]
- name: Template Kubernetes unencrypted files
when: item.state == 'file' and 'sops' not in item.path and '.DS_Store' not in item.path
ansible.builtin.template:
src: "{{ item.src }}"
dest: "{{ repository_path }}/kubernetes/{{ item.path | regex_replace('.j2$', '') }}"
mode: "0755"
with_community.general.filetree: ["../templates/kubernetes/"]
# https://github.com/ansible-collections/community.sops/issues/153
- name: Template Kubernetes encrypted files
block:
- name: Rename Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml"
dest: "{{ repository_path }}/.sops.yaml.orig"
mode: "0755"
- name: Remove Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml"
state: absent
- name: Template Kubernetes encrypted files
when: item.state == 'file' and 'sops' in item.path
community.sops.sops_encrypt:
path: "{{ repository_path }}/kubernetes/{{ item.path | replace('.j2', '') }}"
encrypted_regex: ^(data|stringData)$
age: ["{{ bootstrap_age_public_key }}"]
content_yaml: "{{ lookup('ansible.builtin.template', item.src) | from_yaml }}"
mode: "0755"
force: true
with_community.general.filetree: ["../templates/kubernetes/"]
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
rescue:
- name: Restore Sops configuration file
ansible.builtin.copy:
src: "{{ repository_path }}/.sops.yaml.orig"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
- name: Remove backed up Sops configuration file
ansible.builtin.file:
path: "{{ repository_path }}/.sops.yaml.orig"
state: absent
+6
View File
@@ -0,0 +1,6 @@
---
- name: Template Sops configuration file
ansible.builtin.template:
src: "templates/.sops.yaml.j2"
dest: "{{ repository_path }}/.sops.yaml"
mode: "0755"
+22
View File
@@ -0,0 +1,22 @@
---
- name: Query age key file
ansible.builtin.stat:
path: ~/.config/sops/age/keys.txt
register: result
- name: Check if age key file exists
ansible.builtin.assert:
that: [result.stat.exists]
success_msg: Age file ~/.config/sops/age/keys.txt exists
fail_msg: Age file ~/.config/sops/age/keys.txt does not exist
- name: Query age key file contents
ansible.builtin.set_fact:
age_contents: "{{ lookup('file', '~/.config/sops/age/keys.txt') }}"
- name: Check if age public keys match
ansible.builtin.assert:
that:
- "'{{ bootstrap_age_public_key }}' in age_contents"
success_msg: Age public key {{ bootstrap_age_public_key }} exists
fail_msg: Age public key {{ bootstrap_age_public_key }} does not exist
+9
View File
@@ -0,0 +1,9 @@
---
- name: Check for required CLI tools
ansible.builtin.shell: |
command -v {{ item }} >/dev/null 2>&1
loop: [age, cloudflared, flux, git, pip3, sops, task]
changed_when: false
check_mode: false
register: result
failed_when: result.rc != 0 and result.rc != 127
@@ -0,0 +1,34 @@
---
- name: Query Cloudflare zone
ansible.builtin.uri:
url: "https://api.cloudflare.com/client/v4/zones?name={{ bootstrap_cloudflare_domain }}&status=active"
headers:
Authorization: "Bearer {{ bootstrap_cloudflare_token }}"
Content-Type: application/json
timeout: 5
return_content: true
body_format: json
register: result
- name: Check if Cloudflare zone exists
ansible.builtin.assert:
that: ["result.json.success == true"]
success_msg: "Cloudflare zone {{ bootstrap_cloudflare_domain }} exists"
fail_msg: "Cloudflare zone {{ bootstrap_cloudflare_domain }} does not exist"
- name: Query Cloudflared tunnel
ansible.builtin.uri:
url: "https://api.cloudflare.com/client/v4/accounts/{{ bootstrap_cloudflare_account_tag }}/cfd_tunnel/{{ bootstrap_cloudflare_tunnel_id }}"
headers:
Authorization: "Bearer {{ bootstrap_cloudflare_token }}"
Content-Type: application/json
timeout: 5
return_content: true
body_format: json
register: result
- name: Check if Cloudflared tunnel exists
ansible.builtin.assert:
that: ["result.json.success == true"]
success_msg: "Cloudflared tunnel {{ bootstrap_cloudflare_tunnel_id }} exists"
fail_msg: "Cloudflared tunnel {{ bootstrap_cloudflare_tunnel_id }} does not exist"
+30
View File
@@ -0,0 +1,30 @@
---
- name: Query Github username
ansible.builtin.uri:
url: "https://api.github.com/users/{{ bootstrap_github_username }}"
timeout: 5
return_content: true
body_format: json
register: result
- name: Check if username exists
ansible.builtin.assert:
that:
- "result.json.login == '{{ bootstrap_github_username }}'"
success_msg: "Github user {{ bootstrap_github_username }} exists"
fail_msg: "Github user {{ bootstrap_github_username }} does not exist"
- name: Query Github repo
ansible.builtin.uri:
url: "https://api.github.com/repos/{{ bootstrap_github_username }}/{{ bootstrap_github_repository_name }}"
timeout: 5
return_content: true
body_format: json
register: result
- name: Check if repo exists
ansible.builtin.assert:
that:
- "result.json.full_name == '{{ bootstrap_github_username }}/{{ bootstrap_github_repository_name }}'"
success_msg: "Github repo {{ bootstrap_github_username }}/{{ bootstrap_github_repository_name }} exists"
fail_msg: "Github repo {{ bootstrap_github_username }}/{{ bootstrap_github_repository_name }} does not exist"
+6
View File
@@ -0,0 +1,6 @@
---
- name: Verify configuration
ansible.builtin.include_tasks: "{{ task }}.yaml"
loop: [vars, age, cli, net, cloudflare, github]
loop_control:
loop_var: task
+94
View File
@@ -0,0 +1,94 @@
---
- name: Verify master node count
ansible.builtin.assert:
that:
- "{{ bootstrap_nodes.master | length > 0 }}"
- "{{ bootstrap_nodes.master | length is odd }}"
success_msg: "Master node count {{ bootstrap_nodes.master | length }} is correct."
fail_msg: "Master node count {{ bootstrap_nodes.master | length }} is not greater than 0 or is not odd."
- name: Verify node CIDR
ansible.builtin.assert:
that: ["{{ bootstrap_node_cidr is ansible.utils.ipv4 }}"]
success_msg: "Node CIDR {{ bootstrap_node_cidr }} is valid."
fail_msg: "Node CIDR {{ bootstrap_node_cidr }} is invalid."
- name: Verify cluster CIDR
ansible.builtin.assert:
that: ["{{ bootstrap_cluster_cidr is ansible.utils.ipv4 }}"]
success_msg: "Cluster CIDR {{ bootstrap_cluster_cidr }} is valid."
fail_msg: "Cluster CIDR {{ bootstrap_cluster_cidr }} is invalid."
- name: Verify service CIDR
ansible.builtin.assert:
that: ["{{ bootstrap_service_cidr is ansible.utils.ipv4 }}"]
success_msg: "Service CIDR {{ bootstrap_service_cidr }} is valid."
fail_msg: "Service CIDR {{ bootstrap_service_cidr }} is invalid."
- name: Verify k8s_gateway
ansible.builtin.assert:
that: ["{{ bootstrap_k8s_gateway_addr is ansible.utils.ipv4 }}"]
success_msg: "k8s_gateway address {{ bootstrap_k8s_gateway_addr }} is valid."
fail_msg: "k8s_gateway address {{ bootstrap_k8s_gateway_addr }} is invalid."
- name: Verify k8s_gateway in node CIDR
ansible.builtin.assert:
that: ["{{ bootstrap_node_cidr | ansible.utils.network_in_usable(bootstrap_k8s_gateway_addr) }}"]
success_msg: "k8s_gateway address {{ bootstrap_k8s_gateway_addr }} is within {{ bootstrap_node_cidr }}."
fail_msg: "k8s_gateway address {{ bootstrap_k8s_gateway_addr }} is not within {{ bootstrap_node_cidr }}."
- name: Verify ingress-nginx
ansible.builtin.assert:
that: ["{{ bootstrap_ingress_nginx_addr is ansible.utils.ipv4 }}"]
success_msg: "ingress-nginx address {{ bootstrap_ingress_nginx_addr }} is valid."
fail_msg: "ingress-nginx address {{ bootstrap_ingress_nginx_addr }} is invalid."
- name: Verify ingress-nginx in node CIDR
ansible.builtin.assert:
that: ["{{ bootstrap_node_cidr | ansible.utils.network_in_usable(bootstrap_ingress_nginx_addr) }}"]
success_msg: "ingress-nginx address {{ bootstrap_ingress_nginx_addr }} is within {{ bootstrap_node_cidr }}."
fail_msg: "ingress-nginx address {{ bootstrap_ingress_nginx_addr }} is not within {{ bootstrap_node_cidr }}."
- name: Verify kube-vip
ansible.builtin.assert:
that: ["{{ bootstrap_kube_vip_addr is ansible.utils.ipv4 }}"]
success_msg: "kube-vip address {{ bootstrap_kube_vip_addr }} is valid."
fail_msg: "kube-vip address {{ bootstrap_kube_vip_addr }} is invalid."
- name: Verify kube-vip in node CIDR
ansible.builtin.assert:
that: ["{{ bootstrap_node_cidr | ansible.utils.network_in_usable(bootstrap_kube_vip_addr) }}"]
success_msg: "kube-vip address {{ bootstrap_kube_vip_addr }} is within {{ bootstrap_node_cidr }}."
fail_msg: "kube-vip address {{ bootstrap_kube_vip_addr }} is not within {{ bootstrap_node_cidr }}."
- name: Verify nodes are not the same IPs as k8s_gateway, ingress-nginx or kube-vip
ansible.builtin.assert:
that:
- "{{ item.address not in (bootstrap_k8s_gateway_addr, bootstrap_ingress_nginx_addr, bootstrap_kube_vip_addr) }}"
success_msg: "Node address {{ item.address }} is different than k8s_gateway, ingress-nginx or kube-vip."
fail_msg: "Node address {{ item.address }} is not different than k8s_gateway, ingress-nginx or kube-vip."
quiet: true
loop: "{{ bootstrap_nodes.master + bootstrap_nodes.worker | default([]) }}"
loop_control:
label: "{{ item.address }}"
- name: Verify nodes are in node CIDR
ansible.builtin.assert:
that: ["{{ bootstrap_node_cidr | ansible.utils.network_in_usable(item.address) }}"]
success_msg: "Node address {{ item.address }} is within {{ bootstrap_node_cidr }}."
fail_msg: "Node address {{ item.address }} is not within {{ bootstrap_node_cidr }}."
quiet: true
loop: "{{ bootstrap_nodes.master + bootstrap_nodes.worker | default([]) }}"
loop_control:
label: "{{ item.address }}"
- name: Verify SSH port is reachable
ansible.builtin.wait_for:
host: "{{ item.address }}"
port: 22
search_regex: OpenSSH
timeout: 10
connection: local
loop: "{{ bootstrap_nodes.master + bootstrap_nodes.worker | default([]) }}"
loop_control:
label: "{{ item.address }}"
+36
View File
@@ -0,0 +1,36 @@
---
- name: Verify required bootstrap vars are set
ansible.builtin.assert:
that:
- "{{ item }} | default('', true) | trim != ''"
success_msg: "Required bootstrap var {{ item }} exists and is defined"
fail_msg: "Required bootstrap var {{ item }} does not exists or is not defined"
loop:
- bootstrap_github_username
- bootstrap_github_repository_name
- bootstrap_age_public_key
- bootstrap_timezone
- bootstrap_acme_email
- bootstrap_flux_github_webhook_token
- bootstrap_cloudflare_domain
- bootstrap_cloudflare_token
- bootstrap_cloudflare_account_tag
- bootstrap_cloudflare_tunnel_secret
- bootstrap_cloudflare_tunnel_id
- bootstrap_node_cidr
- bootstrap_k8s_gateway_addr
- bootstrap_ingress_nginx_addr
- bootstrap_kube_vip_addr
- bootstrap_cluster_cidr
- bootstrap_service_cidr
- name: Verify bootstrap node names are valid
ansible.builtin.assert:
that:
- "item.name is match('^[a-z0-9-]+$')"
success_msg: "Node name {{ item.name }} is valid"
fail_msg: "Node name {{ item.name }} is not valid"
loop: "{{ bootstrap_nodes.master + bootstrap_nodes.worker | default([]) }}"
loop_control:
label: "{{ item.name }}"
+11
View File
@@ -0,0 +1,11 @@
---
creation_rules:
- path_regex: kubernetes/.*\.sops\.ya?ml
encrypted_regex: "^(data|stringData)$"
key_groups:
- age:
- "{{ bootstrap_age_public_key }}"
- path_regex: ansible/.*\.sops\.ya?ml
key_groups:
- age:
- "{{ bootstrap_age_public_key }}"
@@ -0,0 +1,30 @@
---
apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
name: csi-driver-nfs
namespace: kube-system
spec:
interval: 30m
chart:
spec:
chart: csi-driver-nfs
version: v4.4.0
sourceRef:
kind: HelmRepository
name: csi-driver-nfs
namespace: flux-system
maxHistory: 2
install:
createNamespace: true
remediation:
retries: 3
upgrade:
cleanupOnFail: true
remediation:
retries: 3
uninstall:
keepHistory: false
values:
externalSnapshotter:
enabled: false
@@ -0,0 +1,7 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
namespace: kube-system
resources:
- ./helmrelease.yaml
- ./storageclass.yaml
@@ -0,0 +1,15 @@
#jinja2: trim_blocks: True, lstrip_blocks: True
{% for item in csi_driver_nfs.storage_class %}
---
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
name: {{ item.name }}
provisioner: nfs.csi.k8s.io
parameters:
server: {{ item.server }}
share: {{ item.share }}
reclaimPolicy: Delete
volumeBindingMode: Immediate
mountOptions: ["hard", "noatime"]
{% endfor %}
@@ -0,0 +1,16 @@
---
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: cluster-apps-csi-driver-nfs
namespace: flux-system
spec:
path: ./kubernetes/apps/kube-system/csi-driver-nfs/app
prune: true
sourceRef:
kind: GitRepository
name: home-kubernetes
wait: true
interval: 30m
retryInterval: 1m
timeout: 5m
@@ -0,0 +1,100 @@
---
apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
name: grafana
namespace: monitoring
spec:
interval: 30m
chart:
spec:
chart: grafana
version: 6.57.4
sourceRef:
kind: HelmRepository
name: grafana
namespace: flux-system
maxHistory: 2
install:
createNamespace: true
remediation:
retries: 3
upgrade:
cleanupOnFail: true
remediation:
retries: 3
uninstall:
keepHistory: false
dependsOn:
- name: local-path-provisioner
namespace: kube-system
- name: ingress-nginx
namespace: networking
values:
deploymentStrategy:
type: Recreate
admin:
existingSecret: grafana-admin-secret
env:
GF_EXPLORE_ENABLED: true
GF_SERVER_ROOT_URL: "https://grafana.${SECRET_DOMAIN}"
grafana.ini:
analytics:
check_for_updates: false
check_for_plugin_updates: false
dashboardProviders:
dashboardproviders.yaml:
apiVersion: 1
providers:
- name: default
orgId: 1
folder: ""
type: file
disableDeletion: false
editable: true
options:
path: /var/lib/grafana/dashboards/default
datasources:
datasources.yaml:
apiVersion: 1
datasources:
- name: Prometheus
type: prometheus
access: proxy
url: http://kube-prometheus-stack-prometheus.monitoring.svc.cluster.local:9090
isDefault: true
dashboards:
default:
cert-manager:
url: https://raw.githubusercontent.com/monitoring-mixins/website/master/assets/cert-manager/dashboards/cert-manager.json
datasource: Prometheus
flux-cluster:
url: https://raw.githubusercontent.com/fluxcd/flux2/main/manifests/monitoring/monitoring-config/dashboards/cluster.json
datasource: Prometheus
flux-control-plane:
url: https://raw.githubusercontent.com/fluxcd/flux2/main/manifests/monitoring/monitoring-config/dashboards/control-plane.json
datasource: Prometheus
sidecar:
dashboards:
enabled: true
searchNamespace: ALL
datasources:
enabled: true
searchNamespace: ALL
serviceMonitor:
enabled: true
ingress:
enabled: true
ingressClassName: nginx
annotations:
hajimari.io/icon: simple-icons:grafana
hosts:
- &host "grafana.${SECRET_DOMAIN}"
tls:
- hosts:
- *host
persistence:
enabled: true
storageClassName: local-path
testFramework:
enabled: false
@@ -0,0 +1,7 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
namespace: monitoring
resources:
- ./secret.sops.yaml
- ./helmrelease.yaml
@@ -0,0 +1,9 @@
---
apiVersion: v1
kind: Secret
metadata:
name: grafana-admin-secret
namespace: monitoring
stringData:
admin-user: admin
admin-password: "{{ grafana.password }}"
@@ -0,0 +1,16 @@
---
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: cluster-apps-grafana
namespace: flux-system
spec:
path: ./kubernetes/apps/monitoring/grafana/app
prune: true
sourceRef:
kind: GitRepository
name: home-kubernetes
wait: false # no flux ks dependents
interval: 30m
retryInterval: 1m
timeout: 5m
@@ -0,0 +1,70 @@
---
apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
name: hajimari
namespace: default
spec:
interval: 30m
chart:
spec:
chart: hajimari
version: 2.0.2
sourceRef:
kind: HelmRepository
name: hajimari
namespace: flux-system
maxHistory: 2
install:
createNamespace: true
remediation:
retries: 3
upgrade:
cleanupOnFail: true
remediation:
retries: 3
uninstall:
keepHistory: false
dependsOn:
- name: ingress-nginx
namespace: networking
values:
hajimari:
title: Apps
darkTheme: espresso
alwaysTargetBlank: true
showGreeting: false
showAppGroups: false
showAppStatus: false
showBookmarkGroups: false
showGlobalBookmarks: false
showAppUrls: false
defaultEnable: true
namespaceSelector:
matchNames:
- default
- monitoring
ingress:
main:
enabled: true
ingressClassName: nginx
annotations:
hajimari.io/enable: "false"
hosts:
- host: &host "hajimari.${SECRET_DOMAIN}"
paths:
- path: /
pathType: Prefix
tls:
- hosts:
- *host
podAnnotations:
configmap.reloader.stakater.com/reload: hajimari-settings
persistence:
data:
enabled: true
type: emptyDir
resources:
requests:
cpu: 100m
memory: 128M
@@ -0,0 +1,6 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
namespace: default
resources:
- ./helmrelease.yaml
@@ -0,0 +1,16 @@
---
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: cluster-apps-hajimari
namespace: flux-system
spec:
path: ./kubernetes/apps/default/hajimari/app
prune: true
sourceRef:
kind: GitRepository
name: home-kubernetes
wait: false # no flux ks dependents
interval: 30m
retryInterval: 1m
timeout: 5m
@@ -0,0 +1,142 @@
#jinja2: trim_blocks: True, lstrip_blocks: True
---
apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
name: kube-prometheus-stack
namespace: monitoring
spec:
interval: 30m
timeout: 15m
chart:
spec:
chart: kube-prometheus-stack
version: 47.1.0
sourceRef:
kind: HelmRepository
name: prometheus-community
namespace: flux-system
maxHistory: 2
install:
createNamespace: true
crds: CreateReplace
remediation:
retries: 3
upgrade:
cleanupOnFail: true
crds: CreateReplace
remediation:
retries: 3
uninstall:
keepHistory: false
dependsOn:
- name: local-path-provisioner
namespace: kube-system
- name: ingress-nginx
namespace: networking
values:
cleanPrometheusOperatorObjectNames: true
alertmanager:
enabled: false
grafana:
enabled: false
forceDeployDashboards: true
sidecar:
dashboards:
multicluster:
etcd:
enabled: true
kube-state-metrics:
metricLabelsAllowlist:
- "pods=[*]"
- "deployments=[*]"
- "persistentvolumeclaims=[*]"
prometheus:
monitor:
enabled: true
relabelings:
- action: replace
regex: (.*)
replacement: $1
sourceLabels: ["__meta_kubernetes_pod_node_name"]
targetLabel: kubernetes_node
kubelet:
enabled: true
serviceMonitor:
metricRelabelings:
# Remove duplicate metrics
- sourceLabels: ["__name__"]
regex: "(apiserver_audit|apiserver_client|apiserver_delegated|apiserver_envelope|apiserver_storage|apiserver_webhooks|authentication_token|cadvisor_version|container_blkio|container_cpu|container_fs|container_last|container_memory|container_network|container_oom|container_processes|container|csi_operations|disabled_metric|get_token|go|hidden_metric|kubelet_certificate|kubelet_cgroup|kubelet_container|kubelet_containers|kubelet_cpu|kubelet_device|kubelet_graceful|kubelet_http|kubelet_lifecycle|kubelet_managed|kubelet_node|kubelet_pleg|kubelet_pod|kubelet_run|kubelet_running|kubelet_runtime|kubelet_server|kubelet_started|kubelet_volume|kubernetes_build|kubernetes_feature|machine_cpu|machine_memory|machine_nvm|machine_scrape|node_namespace|plugin_manager|prober_probe|process_cpu|process_max|process_open|process_resident|process_start|process_virtual|registered_metric|rest_client|scrape_duration|scrape_samples|scrape_series|storage_operation|volume_manager|volume_operation|workqueue)_(.+)"
action: keep
- sourceLabels: ["node"]
targetLabel: instance
action: replace
kubeApiServer:
enabled: true
serviceMonitor:
metricRelabelings:
# Remove duplicate metrics
- sourceLabels: ["__name__"]
regex: "(aggregator_openapi|aggregator_unavailable|apiextensions_openapi|apiserver_admission|apiserver_audit|apiserver_cache|apiserver_cel|apiserver_client|apiserver_crd|apiserver_current|apiserver_envelope|apiserver_flowcontrol|apiserver_init|apiserver_kube|apiserver_longrunning|apiserver_request|apiserver_requested|apiserver_response|apiserver_selfrequest|apiserver_storage|apiserver_terminated|apiserver_tls|apiserver_watch|apiserver_webhooks|authenticated_user|authentication|disabled_metric|etcd_bookmark|etcd_lease|etcd_request|field_validation|get_token|go|grpc_client|hidden_metric|kube_apiserver|kubernetes_build|kubernetes_feature|node_authorizer|pod_security|process_cpu|process_max|process_open|process_resident|process_start|process_virtual|registered_metric|rest_client|scrape_duration|scrape_samples|scrape_series|serviceaccount_legacy|serviceaccount_stale|serviceaccount_valid|watch_cache|workqueue)_(.+)"
action: keep
# Remove high cardinality metrics
- sourceLabels: ["__name__"]
regex: (apiserver|etcd|rest_client)_request(|_sli|_slo)_duration_seconds_bucket
action: drop
- sourceLabels: ["__name__"]
regex: (apiserver_response_sizes_bucket|apiserver_watch_events_sizes_bucket)
action: drop
kubeControllerManager:
enabled: true
endpoints: &cp
{% for item in bootstrap_nodes.master %}
- {{ item.address }}
{% endfor %}
serviceMonitor:
metricRelabelings:
# Remove duplicate metrics
- sourceLabels: ["__name__"]
regex: "(apiserver_audit|apiserver_client|apiserver_delegated|apiserver_envelope|apiserver_storage|apiserver_webhooks|attachdetach_controller|authenticated_user|authentication|cronjob_controller|disabled_metric|endpoint_slice|ephemeral_volume|garbagecollector_controller|get_token|go|hidden_metric|job_controller|kubernetes_build|kubernetes_feature|leader_election|node_collector|node_ipam|process_cpu|process_max|process_open|process_resident|process_start|process_virtual|pv_collector|registered_metric|replicaset_controller|rest_client|retroactive_storageclass|root_ca|running_managed|scrape_duration|scrape_samples|scrape_series|service_controller|storage_count|storage_operation|ttl_after|volume_operation|workqueue)_(.+)"
action: keep
kubeEtcd:
enabled: true
endpoints: *cp
kubeProxy:
enabled: false # Disabled because eBPF
kubeScheduler:
enabled: true
endpoints: *cp
serviceMonitor:
metricRelabelings:
# Remove duplicate metrics
- sourceLabels: ["__name__"]
regex: "(apiserver_audit|apiserver_client|apiserver_delegated|apiserver_envelope|apiserver_storage|apiserver_webhooks|authenticated_user|authentication|disabled_metric|go|hidden_metric|kubernetes_build|kubernetes_feature|leader_election|process_cpu|process_max|process_open|process_resident|process_start|process_virtual|registered_metric|rest_client|scheduler|scrape_duration|scrape_samples|scrape_series|workqueue)_(.+)"
action: keep
prometheus:
ingress:
enabled: true
ingressClassName: nginx
annotations:
hajimari.io/appName: Prometheus
hajimari.io/icon: simple-icons:prometheus
pathType: Prefix
hosts:
- &host "prometheus.${SECRET_DOMAIN}"
tls:
- hosts:
- *host
prometheusSpec:
ruleSelectorNilUsesHelmValues: false
serviceMonitorSelectorNilUsesHelmValues: false
podMonitorSelectorNilUsesHelmValues: false
probeSelectorNilUsesHelmValues: false
enableAdminAPI: true
walCompression: true
retentionSize: 8GB
storageSpec:
volumeClaimTemplate:
spec:
storageClassName: local-path
resources:
requests:
storage: 10Gi
@@ -0,0 +1,6 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
namespace: monitoring
resources:
- ./helmrelease.yaml
@@ -0,0 +1,16 @@
---
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: cluster-apps-kube-prometheus-stack
namespace: flux-system
spec:
path: ./kubernetes/apps/monitoring/kube-prometheus-stack/app
prune: true
sourceRef:
kind: GitRepository
name: home-kubernetes
wait: false # no flux ks dependents
interval: 30m
retryInterval: 1m
timeout: 5m
@@ -0,0 +1,61 @@
---
apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
name: kubernetes-dashboard
namespace: monitoring
spec:
interval: 30m
chart:
spec:
chart: kubernetes-dashboard
version: 7.0.2
sourceRef:
kind: HelmRepository
name: kubernetes-dashboard
namespace: flux-system
maxHistory: 2
install:
createNamespace: true
remediation:
retries: 3
upgrade:
cleanupOnFail: true
remediation:
retries: 3
uninstall:
keepHistory: false
dependsOn:
- name: ingress-nginx
namespace: networking
values:
env:
TZ: "${TIMEZONE}"
api:
containers:
args:
- --enable-insecure-login=true
- --enable-skip-login=true
- --disable-settings-authorizer=true
app:
ingress:
enabled: true
className: nginx
annotations:
hajimari.io/icon: mdi:kubernetes
hosts:
- &host "kubernetes.${SECRET_DOMAIN}"
tls:
- hosts:
- *host
secretName: kubernetes-dashboard-tls
metricsScraper:
enabled: true
serviceMonitor:
enabled: false
cert-manager:
enabled: false
nginx:
enabled: false
metrics-server:
enabled: false
@@ -0,0 +1,7 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
namespace: monitoring
resources:
- ./rbac.yaml
- ./helmrelease.yaml
@@ -0,0 +1,41 @@
# For dashboard sign in token:
# kubectl -n monitoring get secret kubernetes-dashboard -o jsonpath='{.data.token}' | base64 -d
---
apiVersion: v1
kind: ServiceAccount
metadata:
name: kubernetes-dashboard
namespace: monitoring
labels:
app.kubernetes.io/managed-by: Helm
annotations:
meta.helm.sh/release-name: kubernetes-dashboard
meta.helm.sh/release-namespace: monitoring
secrets:
- name: kubernetes-dashboard
---
apiVersion: v1
kind: Secret
type: kubernetes.io/service-account-token
metadata:
name: kubernetes-dashboard
namespace: monitoring
labels:
app.kubernetes.io/managed-by: Helm
annotations:
meta.helm.sh/release-name: kubernetes-dashboard
meta.helm.sh/release-namespace: monitoring
kubernetes.io/service-account.name: kubernetes-dashboard
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
name: system:kubernetes-dashboard
roleRef:
apiGroup: rbac.authorization.k8s.io
kind: ClusterRole
name: cluster-admin
subjects:
- kind: ServiceAccount
name: kubernetes-dashboard
namespace: monitoring
@@ -0,0 +1,20 @@
---
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: cluster-apps-kubernetes-dashboard
namespace: flux-system
spec:
dependsOn:
- name: cluster-apps-cert-manager
- name: cluster-apps-ingress-nginx
- name: cluster-apps-metrics-server
path: ./kubernetes/apps/monitoring/kubernetes-dashboard/app
prune: true
sourceRef:
kind: GitRepository
name: home-kubernetes
wait: false # no flux ks dependents
interval: 30m
retryInterval: 1m
timeout: 5m
@@ -0,0 +1,54 @@
---
apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
name: weave-gitops
namespace: flux-system
spec:
interval: 30m
chart:
spec:
chart: weave-gitops
version: 4.0.24
sourceRef:
kind: HelmRepository
name: weave-gitops
namespace: flux-system
maxHistory: 2
install:
createNamespace: true
remediation:
retries: 3
upgrade:
cleanupOnFail: true
remediation:
retries: 3
uninstall:
keepHistory: false
values:
adminUser:
create: true
createSecret: false
username: admin
ingress:
enabled: true
className: nginx
annotations:
hajimari.io/icon: sawtooth-wave
hosts:
- host: &host "gitops.${SECRET_DOMAIN}"
paths:
- path: /
pathType: Prefix
tls:
- hosts:
- *host
networkPolicy:
create: false
metrics:
enabled: true
rbac:
create: true
impersonationResourceNames: ["admin"]
podAnnotations:
secret.reloader.stakater.com/reload: cluster-user-auth
@@ -0,0 +1,7 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
namespace: flux-system
resources:
- ./secret.sops.yaml
- ./helmrelease.yaml
@@ -0,0 +1,10 @@
---
apiVersion: v1
kind: Secret
metadata:
name: cluster-user-auth
namespace: flux-system
type: Opaque
stringData:
username: admin
password: "{{ weave_gitops.password | password_hash('bcrypt', rounds=10) }}"
@@ -0,0 +1,16 @@
---
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: cluster-apps-weave-gitops
namespace: flux-system
spec:
path: ./kubernetes/apps/flux-system/weave-gitops/app
prune: true
sourceRef:
kind: GitRepository
name: home-kubernetes
wait: false # no flux ks dependents
interval: 30m
retryInterval: 1m
timeout: 5m
@@ -0,0 +1,35 @@
[defaults]
# General settings
nocows = True
executable = /bin/bash
stdout_callback = yaml
force_valid_group_names = ignore
# File/Directory settings
log_path = ~/.ansible/ansible.log
inventory = ./inventory
roles_path = ~/.ansible/roles:./roles
collections_path = ~/.ansible/collections
remote_tmp = ~/.ansible/tmp
local_tmp = ~/.ansible/tmp
# Fact Caching settings
fact_caching = jsonfile
fact_caching_connection = ~/.ansible/facts_cache
# SSH settings
remote_port = 22
timeout = 60
host_key_checking = False
# Plugin settings
vars_plugins_enabled = host_group_vars,community.sops.sops
[inventory]
unparsed_is_failed = True
[privilege_escalation]
become = True
[ssh_connection]
scp_if_ssh = smart
retries = 3
ssh_args = -o ControlMaster=auto -o ControlPersist=30m -o Compression=yes -o ServerAliveInterval=15s
pipelining = True
control_path = %(directory)s/%%h-%%r
@@ -0,0 +1,32 @@
---
#
# Below vars are for the xanmanning.k3s role
# ...see https://github.com/PyratLabs/ansible-role-k3s
#
# renovate: datasource=github-releases depName=k3s-io/k3s
k3s_release_version: "v1.27.3+k3s1"
k3s_install_hard_links: true
k3s_become: true
k3s_etcd_datastore: true
k3s_use_unsupported_config: true
k3s_registration_address: "{% raw %}{{ kube_vip_addr }}{% endraw %}"
k3s_server_manifests_urls:
# Kube-vip RBAC
- url: https://raw.githubusercontent.com/kube-vip/kube-vip/main/docs/manifests/rbac.yaml
filename: kube-vip-rbac.yaml
# Essential Prometheus Operator CRDs (the rest are installed with the kube-prometheus-stack helm release)
- url: https://raw.githubusercontent.com/prometheus-operator/prometheus-operator/v0.66.0/example/prometheus-operator-crd/monitoring.coreos.com_podmonitors.yaml
filename: custom-prometheus-podmonitors.yaml
- url: https://raw.githubusercontent.com/prometheus-operator/prometheus-operator/v0.66.0/example/prometheus-operator-crd/monitoring.coreos.com_prometheusrules.yaml
filename: custom-prometheus-prometheusrules.yaml
- url: https://raw.githubusercontent.com/prometheus-operator/prometheus-operator/v0.66.0/example/prometheus-operator-crd/monitoring.coreos.com_servicemonitors.yaml
filename: custom-prometheus-servicemonitors.yaml
# /var/lib/rancher/k3s/server/manifests
k3s_server_manifests_templates:
- custom-cilium-helmchart.yaml.j2
- custom-cilium-l2.yaml.j2
- custom-coredns-helmchart.yaml.j2
# /var/lib/rancher/k3s/agent/pod-manifests
k3s_server_pod_manifests_templates:
- kube-vip-static-pod.yaml.j2
@@ -0,0 +1,8 @@
---
timezone: "{{ bootstrap_timezone }}"
github_username: "{{ bootstrap_github_username }}"
coredns_addr: "{{ bootstrap_coredns_addr }}"
kube_vip_addr: "{{ bootstrap_kube_vip_addr }}"
cluster_cidr: "{{ bootstrap_cluster_cidr }}"
service_cidr: "{{ bootstrap_service_cidr }}"
node_cidr: "{{ bootstrap_node_cidr }}"
@@ -0,0 +1,31 @@
---
# https://rancher.com/docs/k3s/latest/en/installation/install-options/server-config/
# https://github.com/PyratLabs/ansible-role-k3s
k3s_control_node: true
k3s_server:
node-ip: "{% raw %}{{ ansible_host }}{% endraw %}"
tls-san:
- "{% raw %}{{ kube_vip_addr }}{% endraw %}"
docker: false
flannel-backend: "none" # This needs to be in quotes
disable:
- coredns # Disable coredns - replaced with Coredns Helm Chart
- flannel # Disable flannel - replaced with Cilium Helm Chart
- local-storage # Disable local-path-provisioner - installed with Flux
- metrics-server # Disable metrics-server - installed with Flux
- servicelb # Disable servicelb - replaced with Cilium Helm Chart
- traefik # Disable traefik - replaced with ingress-nginx and installed with Flux
disable-network-policy: true
disable-cloud-controller: true
disable-kube-proxy: true # Cilium uses eBPF
write-kubeconfig-mode: "644"
cluster-cidr: "{% raw %}{{ cluster_cidr }}{% endraw %}"
service-cidr: "{% raw %}{{ service_cidr }}{% endraw %}"
etcd-expose-metrics: true # Required to monitor etcd with kube-prometheus-stack
kube-controller-manager-arg:
- "bind-address=0.0.0.0" # Required to monitor kube-controller-manager with kube-prometheus-stack
kube-scheduler-arg:
- "bind-address=0.0.0.0" # Required to monitor kube-scheduler with kube-prometheus-stack
kube-apiserver-arg:
- "anonymous-auth=true" # Required for HAProxy health-checks
@@ -0,0 +1,7 @@
---
# https://rancher.com/docs/k3s/latest/en/installation/install-options/agent-config/
# https://github.com/PyratLabs/ansible-role-k3s
k3s_control_node: false
k3s_agent:
node-ip: "{% raw %}{{ ansible_host }}{% endraw %}"
Whitespace-only changes.
@@ -0,0 +1,20 @@
#jinja2: trim_blocks: True, lstrip_blocks: True
---
kubernetes:
children:
master:
hosts:
{% for item in bootstrap_nodes.master %}
{{ item.name }}:
ansible_user: {{ item.username }}
ansible_host: {{ item.address }}
{% endfor %}
{% if bootstrap_nodes.worker | default([]) | length > 0 %}
worker:
hosts:
{% for item in bootstrap_nodes.worker %}
{{ item.name }}:
ansible_user: {{ item.username }}
ansible_host: {{ item.address }}
{% endfor %}
{% endif %}
@@ -0,0 +1,71 @@
---
- name: Cluster Installation
hosts: all
become: true
gather_facts: true
any_errors_fatal: true
pre_tasks:
- name: Pausing for 5 seconds...
ansible.builtin.pause:
seconds: 5
tasks:
- name: Check if cluster is installed
check_mode: false
ansible.builtin.stat:
path: /etc/rancher/k3s/config.yaml
register: k3s_installed
- name: Ignore manifests templates and urls if the cluster is already installed
when: k3s_installed.stat.exists
ansible.builtin.set_fact:
k3s_server_manifests_templates: []
k3s_server_manifests_urls: []
- name: Install Kubernetes
ansible.builtin.include_role:
name: xanmanning.k3s
public: true
vars:
k3s_state: installed
- name: Kubeconfig
ansible.builtin.include_tasks: tasks/kubeconfig.yaml
- name: Wait for custom manifests to rollout
when:
- k3s_primary_control_node
- (k3s_server_manifests_templates | length > 0
or k3s_server_manifests_urls | length > 0)
kubernetes.core.k8s_info:
kubeconfig: /etc/rancher/k3s/k3s.yaml
kind: "{% raw %}{{ item.kind }}{% endraw %}"
name: "{% raw %}{{ item.name }}{% endraw %}"
namespace: "{% raw %}{{ item.namespace | default('') }}{% endraw %}"
wait: true
wait_sleep: 10
wait_timeout: 360
loop:
- { name: cilium, kind: HelmChart, namespace: kube-system }
- { name: coredns, kind: HelmChart, namespace: kube-system }
- { name: policy, kind: CiliumL2AnnouncementPolicy }
- { name: pool, kind: CiliumLoadBalancerIPPool }
- { name: podmonitors.monitoring.coreos.com, kind: CustomResourceDefinition }
- { name: prometheusrules.monitoring.coreos.com, kind: CustomResourceDefinition }
- { name: servicemonitors.monitoring.coreos.com, kind: CustomResourceDefinition }
- name: Coredns
when: k3s_primary_control_node
ansible.builtin.include_tasks: tasks/coredns.yaml
- name: Cilium
when: k3s_primary_control_node
ansible.builtin.include_tasks: tasks/cilium.yaml
- name: Cruft
when: k3s_primary_control_node
ansible.builtin.include_tasks: tasks/cruft.yaml
- name: Stale Containers
ansible.builtin.include_tasks: tasks/stale_containers.yaml
vars:
stale_containers_state: enabled
@@ -0,0 +1,23 @@
---
- name: Cluster kube-vip
hosts: master
become: true
gather_facts: true
any_errors_fatal: true
pre_tasks:
- name: Pausing for 5 seconds...
ansible.builtin.pause:
seconds: 5
tasks:
- name: Ensure Kubernetes is running
ansible.builtin.include_role:
name: xanmanning.k3s
public: true
vars:
k3s_state: started
- name: Upgrade kube-vip
ansible.builtin.template:
src: templates/kube-vip-static-pod.yaml.j2
dest: "{% raw %}{{ k3s_server_pod_manifests_dir }}{% endraw %}/kube-vip-static-pod.yaml"
mode: preserve
@@ -0,0 +1,73 @@
---
- name: Cluster Nuke
hosts: all
become: true
gather_facts: true
any_errors_fatal: true
vars_prompt:
- name: nuke
prompt: |-
Are you sure you want to nuke this cluster?
Type 'YES I WANT TO DESTROY THIS CLUSTER' to proceed
default: "n"
private: false
pre_tasks:
- name: Check for confirmation
ansible.builtin.fail:
msg: Aborted nuking the cluster
when: nuke != 'YES I WANT TO DESTROY THIS CLUSTER'
- name: Pausing for 5 seconds...
ansible.builtin.pause:
seconds: 5
tasks:
- name: Stop Kubernetes # noqa: ignore-errors
ignore_errors: true
block:
- name: Stop Kubernetes
ansible.builtin.include_role:
name: xanmanning.k3s
public: true
vars:
k3s_state: stopped
# https://github.com/k3s-io/docs/blob/main/docs/installation/network-options.md
- name: Networking
block:
- name: Networking | Delete Cilium links
ansible.builtin.command:
cmd: "ip link delete {% raw %}{{ item }}{% endraw %}"
removes: "/sys/class/net/{% raw %}{{ item }}{% endraw %}"
loop: ["cilium_host", "cilium_net", "cilium_vxlan"]
- name: Networking | Flush iptables
ansible.builtin.iptables:
table: "{% raw %}{{ item }}{% endraw %}"
flush: true
loop: ["filter", "nat", "mangle", "raw"]
- name: Networking | Flush ip6tables
ansible.builtin.iptables:
table: "{% raw %}{{ item }}{% endraw %}"
flush: true
ip_version: ipv6
loop: ["filter", "nat", "mangle", "raw"]
- name: Networking | Delete CNI directory
ansible.builtin.file:
path: /etc/cni/net.d
state: absent
- name: Uninstall Kubernetes
ansible.builtin.include_role:
name: xanmanning.k3s
public: true
vars:
k3s_state: uninstalled
- name: Stale Containers
ansible.builtin.include_tasks: tasks/stale_containers.yaml
vars:
stale_containers_state: disabled
- name: Reboot
ansible.builtin.reboot:
msg: Rebooting nodes
reboot_timeout: 3600
@@ -0,0 +1,120 @@
---
- name: Prepare System
hosts: all
become: true
gather_facts: true
any_errors_fatal: true
pre_tasks:
- name: Pausing for 5 seconds...
ansible.builtin.pause:
seconds: 5
tasks:
- name: Locale
block:
- name: Locale | Set timezone
community.general.timezone:
name: "{% raw %}{{ timezone | default('Etc/UTC') }}{% endraw %}"
- name: Packages
block:
- name: Packages | Install
ansible.builtin.apt:
name: apt-transport-https,ca-certificates,conntrack,curl,dirmngr,gdisk,gnupg,htop,
iptables,iputils-ping,ipvsadm,libseccomp2,lm-sensors,neofetch,net-tools,nfs-common,
nvme-cli,open-iscsi,parted,psmisc,python3,python3-apt,python3-kubernetes,python3-yaml,
smartmontools,socat,software-properties-common,unzip,util-linux
install_recommends: false
- name: User Configuration
block:
- name: User Configuration | SSH keys
ansible.posix.authorized_key:
user: "{% raw %}{{ ansible_user }}{% endraw %}"
key: "https://github.com/{% raw %}{{ github_username }}{% endraw %}.keys"
- name: User Configuration | Silence login
ansible.builtin.file:
dest: /root/.hushlogin
state: touch
owner: "{% raw %}{{ ansible_user }}{% endraw %}"
mode: "0644"
modification_time: preserve
access_time: preserve
- name: Network Configuration
notify: Reboot
block:
- name: Network Configuration | Set hostname
ansible.builtin.hostname:
name: "{% raw %}{{ inventory_hostname }}{% endraw %}"
- name: Network Configuration | Update hosts
ansible.builtin.copy:
content: |
127.0.0.1 localhost
127.0.1.1 {% raw %}{{ inventory_hostname }}{% endraw %}
# The following lines are desirable for IPv6 capable hosts
::1 localhost ip6-localhost ip6-loopback
ff02::1 ip6-allnodes
ff02::2 ip6-allrouters
dest: /etc/hosts
mode: preserve
# https://github.com/cilium/cilium/issues/18706
- name: Network Configuration | Cilium (1)
ansible.builtin.lineinfile:
dest: /etc/systemd/networkd.conf
regexp: ManageForeignRoutingPolicyRules
line: ManageForeignRoutingPolicyRules=no
- name: Network Configuration | Cilium (2)
ansible.builtin.lineinfile:
dest: /etc/systemd/networkd.conf
regexp: ManageForeignRoutes
line: ManageForeignRoutes=no
- name: System Configuration
notify: Reboot
block:
- name: System Configuration | Neofetch
ansible.builtin.copy:
dest: /etc/profile.d/neofetch.sh
mode: "0755"
content: neofetch --config none
- name: System Configuration | Disable apparmor
ansible.builtin.systemd:
name: apparmor
state: stopped
masked: true
- name: System Configuration | Disable swap
ansible.posix.mount:
name: "{% raw %}{{ item }}{% endraw %}"
fstype: swap
state: absent
loop: ["none", "swap"]
- name: System Configuration | Kernel modules (1)
community.general.modprobe:
name: "{% raw %}{{ item }}{% endraw %}"
state: present
loop: ["br_netfilter", "ceph", "ip_vs", "ip_vs_rr", "nbd", "overlay", "rbd"]
- name: System Configuration | Kernel modules (2)
ansible.builtin.copy:
dest: "/etc/modules-load.d/{% raw %}{{ item }}{% endraw %}.conf"
mode: "0644"
content: "{% raw %}{{ item }}{% endraw %}"
loop: ["br_netfilter", "ceph", "ip_vs", "ip_vs_rr", "nbd", "overlay", "rbd"]
- name: System Configuration | Sysctl
ansible.posix.sysctl:
name: "{% raw %}{{ item.key }}{% endraw %}"
value: "{% raw %}{{ item.value }}{% endraw %}"
sysctl_file: /etc/sysctl.d/99-kubernetes.conf
reload: true
with_dict: "{% raw %}{{ sysctl_config }}{% endraw %}"
vars:
sysctl_config:
fs.inotify.max_queued_events: 65536
fs.inotify.max_user_watches: 524288
fs.inotify.max_user_instances: 8192
handlers:
- name: Reboot
ansible.builtin.reboot:
msg: Rebooting nodes
reboot_timeout: 3600
@@ -0,0 +1,15 @@
---
- name: Reboot
hosts: all
become: true
gather_facts: true
any_errors_fatal: true
pre_tasks:
- name: Pausing for 5 seconds...
ansible.builtin.pause:
seconds: 5
tasks:
- name: Reboot
ansible.builtin.reboot:
msg: Rebooting nodes
reboot_timeout: 3600
@@ -0,0 +1,70 @@
---
# https://github.com/kevincoakley/ansible-role-k8s-rolling-update
- name: Cluster rollout update
hosts: all
become: true
gather_facts: true
any_errors_fatal: true
serial: 1
pre_tasks:
- name: Pausing for 5 seconds...
ansible.builtin.pause:
seconds: 5
tasks:
- name: Get the node's details
ansible.builtin.command: "kubectl get node {% raw %}{{ inventory_hostname }}{% endraw %} -o json"
register: kubectl_get_node
delegate_to: "{% raw %}{{ groups['master'][0] }}{% endraw %}"
failed_when: false
changed_when: false
- name: Update node
when:
# When status.conditions[x].type == Ready then check stats.conditions[x].status for True|False
- kubectl_get_node['stdout'] | from_json | json_query("status.conditions[?type == 'Ready'].status")
# If spec.unschedulable is defined then the node is cordoned
- not (kubectl_get_node['stdout'] | from_json).spec.unschedulable is defined
block:
- name: Cordon node
ansible.builtin.command: "kubectl cordon {% raw %}{{ inventory_hostname }}{% endraw %}"
delegate_to: "{% raw %}{{ groups['master'][0] }}{% endraw %}"
changed_when: false
- name: Wait for node to cordon
ansible.builtin.command: "kubectl get node {% raw %}{{ inventory_hostname }}{% endraw %} -o json"
register: wait_for_cordon
retries: 10
delay: 10
delegate_to: "{% raw %}{{ groups['master'][0] }}{% endraw %}"
changed_when: false
until: (wait_for_cordon['stdout'] | from_json).spec.unschedulable
- name: Cordon & drain node
ansible.builtin.command: "kubectl drain --ignore-daemonsets --delete-emptydir-data --force {% raw %}{{ inventory_hostname }}{% endraw %}"
delegate_to: "{% raw %}{{ groups['master'][0] }}{% endraw %}"
changed_when: false
- name: Update all packages
ansible.builtin.apt:
upgrade: dist
update_cache: true
- name: Reboot
ansible.builtin.reboot:
msg: Rebooting node
post_reboot_delay: 60
reboot_timeout: 3600
- name: Uncordon node
ansible.builtin.command: "kubectl uncordon {% raw %}{{ inventory_hostname }}{% endraw %}"
delegate_to: "{% raw %}{{ groups['master'][0] }}{% endraw %}"
changed_when: false
- name: Wait for node to uncordon
ansible.builtin.command: "kubectl get node {% raw %}{{ inventory_hostname }}{% endraw %} -o json"
register: wait_for_uncordon
retries: 10
delay: 10
delegate_to: "{% raw %}{{ groups['master'][0] }}{% endraw %}"
changed_when: false
until: not (kubectl_get_node['stdout'] | from_json).spec.unschedulable is defined
@@ -0,0 +1,6 @@
[Unit]
Description=Stale containers
[Service]
Type=oneshot
ExecStart=/usr/local/bin/k3s crictl rmi --prune
@@ -0,0 +1,11 @@
[Unit]
Description=Stale containers
[Timer]
OnCalendar=weekly
AccuracySec=1h
Persistent=true
RandomizedDelaySec=6000
[Install]
WantedBy=timers.target
@@ -0,0 +1,56 @@
---
- name: Cilium
block:
- name: Cilium | Check if Cilium HelmChart exists
kubernetes.core.k8s_info:
kubeconfig: /etc/rancher/k3s/k3s.yaml
name: cilium
kind: HelmChart
namespace: kube-system
register: cilium_helmchart
- name: Cilium | Wait for Cilium to rollout
when: cilium_helmchart.resources | count > 0
kubernetes.core.k8s_info:
kubeconfig: /etc/rancher/k3s/k3s.yaml
name: helm-install-cilium
kind: Job
namespace: kube-system
wait: true
wait_condition:
type: Complete
status: true
wait_timeout: 360
- name: Cilium | Patch the Cilium HelmChart to unmanage it
when: cilium_helmchart.resources | count > 0
kubernetes.core.k8s_json_patch:
kubeconfig: /etc/rancher/k3s/k3s.yaml
name: cilium
kind: HelmChart
namespace: kube-system
patch:
- op: add
path: /metadata/annotations/helmcharts.helm.cattle.io~1unmanaged
value: "true"
- name: Cilium | Delete the Cilium HelmChart CR
when: cilium_helmchart.resources | count > 0
kubernetes.core.k8s:
kubeconfig: /etc/rancher/k3s/k3s.yaml
name: cilium
kind: HelmChart
namespace: kube-system
state: absent
- name: Cilium | Force delete the Cilium HelmChart
when: cilium_helmchart.resources | count > 0
kubernetes.core.k8s:
kubeconfig: /etc/rancher/k3s/k3s.yaml
name: cilium
kind: HelmChart
namespace: kube-system
state: patched
definition:
metadata:
finalizers: []
@@ -0,0 +1,56 @@
---
- name: Coredns
block:
- name: Coredns | Check if Coredns HelmChart exists
kubernetes.core.k8s_info:
kubeconfig: /etc/rancher/k3s/k3s.yaml
name: coredns
kind: HelmChart
namespace: kube-system
register: coredns_helmchart
- name: Coredns | Wait for Coredns to rollout
when: coredns_helmchart.resources | count > 0
kubernetes.core.k8s_info:
kubeconfig: /etc/rancher/k3s/k3s.yaml
name: helm-install-coredns
kind: Job
namespace: kube-system
wait: true
wait_condition:
type: Complete
status: true
wait_timeout: 360
- name: Coredns | Patch the Coredns HelmChart to unmanage it
when: coredns_helmchart.resources | count > 0
kubernetes.core.k8s_json_patch:
kubeconfig: /etc/rancher/k3s/k3s.yaml
name: coredns
kind: HelmChart
namespace: kube-system
patch:
- op: add
path: /metadata/annotations/helmcharts.helm.cattle.io~1unmanaged
value: "true"
- name: Coredns | Delete the Coredns HelmChart CR
when: coredns_helmchart.resources | count > 0
kubernetes.core.k8s:
kubeconfig: /etc/rancher/k3s/k3s.yaml
name: coredns
kind: HelmChart
namespace: kube-system
state: absent
- name: Coredns | Force delete the Coredns HelmChart
when: coredns_helmchart.resources | count > 0
kubernetes.core.k8s:
kubeconfig: /etc/rancher/k3s/k3s.yaml
name: coredns
kind: HelmChart
namespace: kube-system
state: patched
definition:
metadata:
finalizers: []
@@ -0,0 +1,32 @@
---
# https://github.com/k3s-io/k3s/issues/1971
- name: Cruft
block:
- name: Cruft | Get list of custom mantifests
ansible.builtin.find:
paths: "{% raw %}{{ k3s_server_manifests_dir }}{% endraw %}"
file_type: file
use_regex: true
patterns: ["^custom-.*"]
register: custom_manifest
- name: Cruft | Delete custom mantifests
ansible.builtin.file:
path: "{% raw %}{{ item.path }}{% endraw %}"
state: absent
loop: "{% raw %}{{ custom_manifest.files }}{% endraw %}"
- name: Cruft | Get list of custom addons
kubernetes.core.k8s_info:
kubeconfig: /etc/rancher/k3s/k3s.yaml
kind: Addon
register: addons_list
- name: Cruft | Delete addons
kubernetes.core.k8s:
kubeconfig: /etc/rancher/k3s/k3s.yaml
name: "{% raw %}{{ item.metadata.name }}{% endraw %}"
kind: Addon
namespace: kube-system
state: absent
loop: "{% raw %}{{ addons_list.resources | selectattr('metadata.name', 'match', '^custom-.*') | list }}{% endraw %}"
@@ -0,0 +1,26 @@
---
- name: Get absolute path to this Git repository # noqa: command-instead-of-module
ansible.builtin.command: git rev-parse --show-toplevel
delegate_to: localhost
become: false
run_once: true
register: repository_path
changed_when: false
check_mode: false
failed_when: repository_path.rc != 0
- name: Copy kubeconfig to the project directory
when: k3s_primary_control_node
ansible.builtin.fetch:
src: /etc/rancher/k3s/k3s.yaml
dest: "{% raw %}{{ repository_path.stdout }}{% endraw %}/kubeconfig"
flat: true
- name: Update kubeconfig with the correct load balancer address
delegate_to: localhost
become: false
run_once: true
ansible.builtin.replace:
path: "{% raw %}{{ repository_path.stdout }}{% endraw %}/kubeconfig"
regexp: https://127.0.0.1:6443
replace: "https://{% raw %}{{ k3s_registration_address }}{% endraw %}:6443"
@@ -0,0 +1,36 @@
---
# https://github.com/k3s-io/k3s/issues/1900
- name: Enabled Stale containers
when: stale_containers_state == "enabled"
block:
- name: Stale containers | Create systemd unit
ansible.builtin.copy:
src: files/stale-containers.service
dest: /etc/systemd/system/stale-containers.service
owner: root
group: root
mode: "0644"
- name: Stale containers | Create systemd timer
ansible.builtin.copy:
src: files/stale-containers.timer
dest: /etc/systemd/system/stale-containers.timer
owner: root
group: root
mode: "0644"
- name: Stale containers | Start the systemd timer
ansible.builtin.systemd:
name: stale-containers.timer
enabled: true
daemon_reload: true
masked: false
state: started
- name: Disable Stale containers
when: stale_containers_state == "disabled"
block:
- name: Stale containers | Mask the systemd timer
ansible.builtin.systemd:
name: stale-containers.timer
masked: true
@@ -0,0 +1,54 @@
---
# https://docs.k3s.io/helm
apiVersion: helm.cattle.io/v1
kind: HelmChart
metadata:
name: cilium
namespace: kube-system
spec:
repo: https://helm.cilium.io/
chart: cilium
# renovate: datasource=helm depName=cilium registryUrl=https://helm.cilium.io
version: "1.14.0-rc.0"
targetNamespace: kube-system
bootstrap: true
valuesContent: |-
autoDirectNodeRoutes: true
bpf:
masquerade: true
bgp:
enabled: false
cluster:
name: home-cluster
id: 1
containerRuntime:
integration: containerd
socketPath: /var/run/k3s/containerd/containerd.sock
endpointRoutes:
enabled: true
hubble:
enabled: false
ipam:
mode: kubernetes
ipv4NativeRoutingCIDR: "{% raw %}{{ k3s_server['cluster-cidr'] }}{% endraw %}"
k8sServiceHost: "{% raw %}{{ kube_vip_addr }}{% endraw %}"
k8sServicePort: 6443
kubeProxyReplacement: strict
kubeProxyReplacementHealthzBindAddr: 0.0.0.0:10256
# https://github.com/cilium/cilium/issues/26586
l2announcements:
enabled: true
leaseDuration: 120s
leaseRenewDeadline: 60s
leaseRetryPeriod: 1s
loadBalancer:
algorithm: maglev
mode: dsr
localRedirectPolicy: true
operator:
replicas: 1
rollOutPods: true
rollOutCiliumPods: true
securityContext:
privileged: true
tunnel: disabled
@@ -0,0 +1,22 @@
---
# https://docs.cilium.io/en/latest/network/l2-announcements
apiVersion: cilium.io/v2alpha1
kind: CiliumL2AnnouncementPolicy
metadata:
name: policy
spec:
loadBalancerIPs: true
# NOTE: This might need to be set if you have more than one active NIC on your nodes
# interfaces:
# - ^eno[0-9]+
nodeSelector:
matchLabels:
kubernetes.io/os: linux
---
apiVersion: cilium.io/v2alpha1
kind: CiliumLoadBalancerIPPool
metadata:
name: pool
spec:
cidrs:
- cidr: "{% raw %}{{ node_cidr }}{% endraw %}"
@@ -0,0 +1,78 @@
---
# https://docs.k3s.io/helm
apiVersion: helm.cattle.io/v1
kind: HelmChart
metadata:
name: coredns
namespace: kube-system
spec:
repo: https://coredns.github.io/helm
chart: coredns
# renovate: datasource=helm depName=coredns registryUrl=https://coredns.github.io/helm
version: "1.24.1"
targetNamespace: kube-system
bootstrap: true
valuesContent: |-
fullnameOverride: coredns
replicaCount: 1
k8sAppLabelOverride: kube-dns
service:
name: kube-dns
# Choose the 10th IP address from the start of the service-cidr
clusterIP: "{% raw %}{{ k3s_server['service-cidr'] | ansible.utils.nthhost(10) }}{% endraw %}"
serviceAccount:
create: true
deployment:
annotations:
reloader.stakater.com/auto: "true"
servers:
- zones:
- zone: .
scheme: dns://
use_tcp: true
port: 53
plugins:
- name: log
- name: errors
- name: health
configBlock: |-
lameduck 5s
- name: ready
- name: kubernetes
parameters: cluster.local in-addr.arpa ip6.arpa
configBlock: |-
pods insecure
fallthrough in-addr.arpa ip6.arpa
ttl 30
- name: prometheus
parameters: 0.0.0.0:9153
- name: forward
parameters: . /etc/resolv.conf
- name: cache
parameters: 30
- name: loop
- name: reload
- name: loadbalance
affinity:
nodeAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
nodeSelectorTerms:
- matchExpressions:
- key: node-role.kubernetes.io/control-plane
operator: Exists
tolerations:
- key: CriticalAddonsOnly
operator: Exists
- key: node-role.kubernetes.io/control-plane
operator: Exists
effect: NoSchedule
- key: node-role.kubernetes.io/master
operator: Exists
effect: NoSchedule
topologySpreadConstraints:
- maxSkew: 1
topologyKey: kubernetes.io/hostname
whenUnsatisfiable: DoNotSchedule
labelSelector:
matchLabels:
app.kubernetes.io/instance: coredns
@@ -0,0 +1,57 @@
---
apiVersion: v1
kind: Pod
metadata:
name: kube-vip
namespace: kube-system
labels:
app.kubernetes.io/instance: kube-vip
app.kubernetes.io/name: kube-vip
spec:
containers:
- name: kube-vip
image: ghcr.io/kube-vip/kube-vip:v0.6.0
imagePullPolicy: IfNotPresent
args: ["manager"]
env:
- name: address
value: "{% raw %}{{ kube_vip_addr }}{% endraw %}"
- name: vip_arp
value: "true"
- name: port
value: "6443"
- name: vip_cidr
value: "32"
- name: cp_enable
value: "true"
- name: cp_namespace
value: kube-system
- name: vip_ddns
value: "false"
- name: svc_enable
value: "false"
- name: vip_leaderelection
value: "true"
- name: vip_leaseduration
value: "15"
- name: vip_renewdeadline
value: "10"
- name: vip_retryperiod
value: "2"
- name: prometheus_server
value: :2112
securityContext:
capabilities:
add: ["NET_ADMIN", "NET_RAW"]
volumeMounts:
- mountPath: /etc/kubernetes/admin.conf
name: kubeconfig
hostAliases:
- hostnames:
- kubernetes
ip: 127.0.0.1
hostNetwork: true
volumes:
- name: kubeconfig
hostPath:
path: /etc/rancher/k3s/k3s.yaml
@@ -0,0 +1 @@
openshift==0.13.1
@@ -0,0 +1,16 @@
---
collections:
- name: ansible.posix
version: 1.5.4
- name: ansible.utils
version: 2.10.3
- name: community.general
version: 7.1.0
- name: community.sops
version: 1.6.4
- name: kubernetes.core
version: 2.4.0
roles:
- name: xanmanning.k3s
src: https://github.com/PyratLabs/ansible-role-k3s.git
version: v3.4.2
@@ -0,0 +1,42 @@
---
apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
name: cert-manager
namespace: cert-manager
spec:
interval: 30m
chart:
spec:
chart: cert-manager
version: v1.12.2
sourceRef:
kind: HelmRepository
name: jetstack
namespace: flux-system
maxHistory: 2
install:
createNamespace: true
remediation:
retries: 3
upgrade:
cleanupOnFail: true
remediation:
retries: 3
uninstall:
keepHistory: false
values:
installCRDs: true
extraArgs:
- --dns01-recursive-nameservers=1.1.1.1:53,9.9.9.9:53
- --dns01-recursive-nameservers-only
podDnsPolicy: None
podDnsConfig:
nameservers:
- "1.1.1.1"
- "9.9.9.9"
prometheus:
enabled: true
servicemonitor:
enabled: true
prometheusInstance: monitoring
@@ -0,0 +1,7 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
namespace: cert-manager
resources:
- ./helmrelease.yaml
- ./prometheusrule.yaml
@@ -0,0 +1,62 @@
#jinja2: trim_blocks: True
---
apiVersion: monitoring.coreos.com/v1
kind: PrometheusRule
metadata:
name: cert-manager.rules
namespace: cert-manager
spec:
groups:
- name: cert-manager
rules:
- alert: CertManagerAbsent
expr: |
absent(up{job="cert-manager"})
for: 15m
labels:
severity: critical
annotations:
description: >
New certificates will not be able to be minted, and existing ones can't be renewed until cert-manager is back.
runbook_url: https://gitlab.com/uneeq-oss/cert-manager-mixin/-/blob/master/RUNBOOK.md#certmanagerabsent
summary: "Cert Manager has dissapeared from Prometheus service discovery."
- name: certificates
rules:
- alert: CertManagerCertExpirySoon
expr: |
avg by (exported_namespace, namespace, name) (certmanager_certificate_expiration_timestamp_seconds - time()) < (21 * 24 * 3600)
for: 15m
labels:
severity: warning
annotations:
description: >
The domain that this cert covers will be unavailable after
{% raw %}{{ $value | humanizeDuration }}{% endraw %}. Clients using endpoints that this cert
protects will start to fail in {% raw %}{{ $value | humanizeDuration }}{% endraw %}.
runbook_url: https://gitlab.com/uneeq-oss/cert-manager-mixin/-/blob/master/RUNBOOK.md#certmanagercertexpirysoon
summary: |
The cert {% raw %}{{ $labels.name }}{% endraw %} is {% raw %}{{ $value | humanizeDuration }}{% endraw %} from expiry, it should have renewed over a week ago.
- alert: CertManagerCertNotReady
expr: |
max by (name, exported_namespace, namespace, condition) (certmanager_certificate_ready_status{condition!="True"} == 1)
for: 15m
labels:
severity: critical
annotations:
description: >
This certificate has not been ready to serve traffic for at least
10m. If the cert is being renewed or there is another valid cert, the ingress
controller _may_ be able to serve that instead.
runbook_url: https://gitlab.com/uneeq-oss/cert-manager-mixin/-/blob/master/RUNBOOK.md#certmanagercertnotready
summary: "The cert {% raw %}{{ $labels.name }}{% endraw %} is not ready to serve traffic."
- alert: CertManagerHittingRateLimits
expr: |
sum by (host) (rate(certmanager_http_acme_client_request_count{status="429"}[5m])) > 0
for: 15m
labels:
severity: critical
annotations:
description: >
Depending on the rate limit, cert-manager may be unable to generate certificates for up to a week.
runbook_url: https://gitlab.com/uneeq-oss/cert-manager-mixin/-/blob/master/RUNBOOK.md#certmanagerhittingratelimits
summary: "Cert manager hitting LetsEncrypt rate limits."
@@ -0,0 +1,39 @@
apiVersion: cert-manager.io/v1
kind: ClusterIssuer
metadata:
name: letsencrypt-production
spec:
acme:
server: https://acme-v02.api.letsencrypt.org/directory
email: "${SECRET_ACME_EMAIL}"
privateKeySecretRef:
name: letsencrypt-production
solvers:
- dns01:
cloudflare:
apiTokenSecretRef:
name: cert-manager-secret
key: api-token
selector:
dnsZones:
- "${SECRET_DOMAIN}"
---
apiVersion: cert-manager.io/v1
kind: ClusterIssuer
metadata:
name: letsencrypt-staging
spec:
acme:
server: https://acme-staging-v02.api.letsencrypt.org/directory
email: "${SECRET_ACME_EMAIL}"
privateKeySecretRef:
name: letsencrypt-staging
solvers:
- dns01:
cloudflare:
apiTokenSecretRef:
name: cert-manager-secret
key: api-token
selector:
dnsZones:
- "${SECRET_DOMAIN}"
@@ -0,0 +1,6 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
- ./secret.sops.yaml
- ./issuers.yaml
@@ -0,0 +1,8 @@
---
apiVersion: v1
kind: Secret
metadata:
name: cert-manager-secret
namespace: cert-manager
stringData:
api-token: "{{ bootstrap_cloudflare_token }}"
@@ -0,0 +1,34 @@
---
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: cluster-apps-cert-manager
namespace: flux-system
spec:
path: ./kubernetes/apps/cert-manager/cert-manager/app
prune: true
sourceRef:
kind: GitRepository
name: home-kubernetes
wait: true
interval: 30m
retryInterval: 1m
timeout: 5m
---
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: cluster-apps-cert-manager-issuers
namespace: flux-system
spec:
dependsOn:
- name: cluster-apps-cert-manager
path: ./kubernetes/apps/cert-manager/cert-manager/issuers
prune: true
sourceRef:
kind: GitRepository
name: home-kubernetes
wait: true
interval: 30m
retryInterval: 1m
timeout: 5m
@@ -0,0 +1,6 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
- ./namespace.yaml
- ./cert-manager/ks.yaml
@@ -0,0 +1,7 @@
---
apiVersion: v1
kind: Namespace
metadata:
name: cert-manager
labels:
kustomize.toolkit.fluxcd.io/prune: disabled
@@ -0,0 +1,9 @@
#jinja2: trim_blocks: True, lstrip_blocks: True
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
- ./namespace.yaml
{% if hajimari.enabled | default(false) %}
- ./hajimari/ks.yaml
{% endif %}
@@ -0,0 +1,7 @@
---
apiVersion: v1
kind: Namespace
metadata:
name: default
labels:
kustomize.toolkit.fluxcd.io/prune: disabled
@@ -0,0 +1,16 @@
---
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: cluster-apps-flux-webhooks
namespace: flux-system
spec:
path: ./kubernetes/apps/flux-system/addons/webhooks
prune: true
sourceRef:
kind: GitRepository
name: home-kubernetes
wait: true
interval: 30m
retryInterval: 1m
timeout: 5m
@@ -0,0 +1,25 @@
---
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
name: flux-webhook
namespace: flux-system
annotations:
external-dns.alpha.kubernetes.io/target: "ingress.${SECRET_DOMAIN}"
hajimari.io/enable: "false"
spec:
ingressClassName: nginx
rules:
- host: &host "flux-webhook.${SECRET_DOMAIN}"
http:
paths:
- path: /hook/
pathType: Prefix
backend:
service:
name: webhook-receiver
port:
number: 80
tls:
- hosts:
- *host
@@ -0,0 +1,7 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
- ./secret.sops.yaml
- ./ingress.yaml
- ./receiver.yaml
@@ -0,0 +1,26 @@
---
apiVersion: notification.toolkit.fluxcd.io/v1
kind: Receiver
metadata:
name: github-receiver
namespace: flux-system
spec:
type: github
events:
- ping
- push
secretRef:
name: github-webhook-token-secret
resources:
- apiVersion: source.toolkit.fluxcd.io/v1
kind: GitRepository
name: home-kubernetes
namespace: flux-system
- apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
name: cluster
namespace: flux-system
- apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
name: cluster-apps
namespace: flux-system
@@ -0,0 +1,8 @@
---
apiVersion: v1
kind: Secret
metadata:
name: github-webhook-token-secret
namespace: flux-system
stringData:
token: "{{ bootstrap_flux_github_webhook_token }}"
@@ -0,0 +1,5 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
- ./github
@@ -0,0 +1,10 @@
#jinja2: trim_blocks: True, lstrip_blocks: True
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
- ./namespace.yaml
- ./addons/ks.yaml
{% if weave_gitops.enabled | default(false) %}
- ./weave-gitops/ks.yaml
{% endif %}
@@ -0,0 +1,7 @@
---
apiVersion: v1
kind: Namespace
metadata:
name: flux-system
labels:
kustomize.toolkit.fluxcd.io/prune: disabled
@@ -0,0 +1,22 @@
---
# https://docs.cilium.io/en/latest/network/l2-announcements
apiVersion: cilium.io/v2alpha1
kind: CiliumL2AnnouncementPolicy
metadata:
name: policy
spec:
loadBalancerIPs: true
# NOTE: This might need to be set if you have more than one active NIC on your nodes
# interfaces:
# - ^eno[0-9]+
nodeSelector:
matchLabels:
kubernetes.io/os: linux
---
apiVersion: cilium.io/v2alpha1
kind: CiliumLoadBalancerIPPool
metadata:
name: pool
spec:
cidrs:
- cidr: "{{ bootstrap_node_cidr }}"
@@ -0,0 +1,105 @@
---
apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
name: cilium
namespace: kube-system
spec:
interval: 30m
chart:
spec:
chart: cilium
version: 1.14.0-rc.0
sourceRef:
kind: HelmRepository
name: cilium
namespace: flux-system
maxHistory: 2
install:
createNamespace: true
remediation:
retries: 3
upgrade:
cleanupOnFail: true
remediation:
retries: 3
uninstall:
keepHistory: false
values:
autoDirectNodeRoutes: true
bpf:
masquerade: true
bgp:
enabled: false
cluster:
name: home-cluster
id: 1
containerRuntime:
integration: containerd
socketPath: /var/run/k3s/containerd/containerd.sock
endpointRoutes:
enabled: true
hubble:
enabled: true
metrics:
enabled:
- dns:query
- drop
- tcp
- flow
- port-distribution
- icmp
- http
serviceMonitor:
enabled: true
relay:
enabled: true
rollOutPods: true
prometheus:
serviceMonitor:
enabled: true
ui:
enabled: true
rollOutPods: true
ingress:
enabled: true
className: nginx
annotations:
hajimari.io/icon: simple-icons:cilium
hosts:
- &host "hubble.${SECRET_DOMAIN}"
tls:
- hosts:
- *host
ipam:
mode: kubernetes
k8sServiceHost: "{{ bootstrap_kube_vip_addr }}"
k8sServicePort: 6443
ipv4NativeRoutingCIDR: "{{ bootstrap_cluster_cidr }}"
kubeProxyReplacement: strict
kubeProxyReplacementHealthzBindAddr: 0.0.0.0:10256
# https://github.com/cilium/cilium/issues/26586
l2announcements:
enabled: true
leaseDuration: 120s
leaseRenewDeadline: 60s
leaseRetryPeriod: 1s
loadBalancer:
algorithm: maglev
mode: dsr
localRedirectPolicy: true
operator:
replicas: 1
rollOutPods: true
prometheus:
enabled: true
serviceMonitor:
enabled: true
prometheus:
enabled: true
serviceMonitor:
enabled: true
rollOutCiliumPods: true
securityContext:
privileged: true
tunnel: disabled
@@ -0,0 +1,7 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
namespace: kube-system
resources:
- ./cilium-l2.yaml
- ./helmrelease.yaml
@@ -0,0 +1,16 @@
---
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: cluster-apps-cilium
namespace: flux-system
spec:
path: ./kubernetes/apps/kube-system/cilium/app
prune: true
sourceRef:
kind: GitRepository
name: home-kubernetes
wait: false # no flux ks dependents
interval: 30m
retryInterval: 1m
timeout: 5m
@@ -0,0 +1,90 @@
---
apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
name: coredns
namespace: kube-system
spec:
interval: 30m
chart:
spec:
chart: coredns
version: 1.24.1
sourceRef:
kind: HelmRepository
name: coredns
namespace: flux-system
maxHistory: 2
install:
createNamespace: true
remediation:
retries: 3
upgrade:
cleanupOnFail: true
remediation:
retries: 3
uninstall:
keepHistory: false
values:
fullnameOverride: coredns
replicaCount: 1
k8sAppLabelOverride: kube-dns
service:
name: kube-dns
clusterIP: "{{ bootstrap_coredns_addr }}"
serviceAccount:
create: true
deployment:
annotations:
reloader.stakater.com/auto: "true"
servers:
- zones:
- zone: .
scheme: dns://
use_tcp: true
port: 53
plugins:
- name: log
- name: errors
- name: health
configBlock: |-
lameduck 5s
- name: ready
- name: kubernetes
parameters: cluster.local in-addr.arpa ip6.arpa
configBlock: |-
pods insecure
fallthrough in-addr.arpa ip6.arpa
ttl 30
- name: prometheus
parameters: 0.0.0.0:9153
- name: forward
parameters: . /etc/resolv.conf
- name: cache
parameters: 30
- name: loop
- name: reload
- name: loadbalance
affinity:
nodeAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
nodeSelectorTerms:
- matchExpressions:
- key: node-role.kubernetes.io/control-plane
operator: Exists
tolerations:
- key: CriticalAddonsOnly
operator: Exists
- key: node-role.kubernetes.io/control-plane
operator: Exists
effect: NoSchedule
- key: node-role.kubernetes.io/master
operator: Exists
effect: NoSchedule
topologySpreadConstraints:
- maxSkew: 1
topologyKey: kubernetes.io/hostname
whenUnsatisfiable: DoNotSchedule
labelSelector:
matchLabels:
app.kubernetes.io/instance: coredns
@@ -0,0 +1,6 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
namespace: kube-system
resources:
- ./helmrelease.yaml
@@ -0,0 +1,16 @@
---
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: cluster-apps-coredns
namespace: flux-system
spec:
path: ./kubernetes/apps/kube-system/coredns/app
prune: true
sourceRef:
kind: GitRepository
name: home-kubernetes
wait: false # no flux ks dependents
interval: 30m
retryInterval: 1m
timeout: 5m
@@ -0,0 +1,14 @@
#jinja2: trim_blocks: True, lstrip_blocks: True
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
- ./namespace.yaml
- ./cilium/ks.yaml
- ./coredns/ks.yaml
- ./local-path-provisioner/ks.yaml
- ./metrics-server/ks.yaml
- ./reloader/ks.yaml
{% if csi_driver_nfs.enabled | default(false) %}
- ./csi-driver-nfs/ks.yaml
{% endif %}
@@ -0,0 +1,72 @@
---
apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
name: local-path-provisioner
namespace: kube-system
spec:
interval: 30m
chart:
spec:
chart: ./deploy/chart/local-path-provisioner
sourceRef:
kind: GitRepository
name: local-path-provisioner
namespace: flux-system
maxHistory: 2
install:
createNamespace: true
remediation:
retries: 3
upgrade:
cleanupOnFail: true
remediation:
retries: 3
uninstall:
keepHistory: false
values:
helperImage:
repository: public.ecr.aws/docker/library/busybox
tag: latest
storageClass:
defaultClass: false
nodePathMap:
- node: DEFAULT_PATH_FOR_NON_LISTED_NODES
paths: ["/var/lib/rancher/k3s/storage"]
# NOTE: Do not enable Flux variable substitution on this HelmRelease
configmap:
setup: |-
#!/bin/sh
while getopts "m:s:p:" opt
do
case $opt in
p)
absolutePath=$OPTARG
;;
s)
sizeInBytes=$OPTARG
;;
m)
volMode=$OPTARG
;;
esac
done
mkdir -m 0777 -p ${absolutePath}
chmod 701 ${absolutePath}/..
teardown: |-
#!/bin/sh
while getopts "m:s:p:" opt
do
case $opt in
p)
absolutePath=$OPTARG
;;
s)
sizeInBytes=$OPTARG
;;
m)
volMode=$OPTARG
;;
esac
done
rm -rf ${absolutePath}
@@ -0,0 +1,6 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
namespace: kube-system
resources:
- ./helmrelease.yaml
@@ -0,0 +1,18 @@
---
apiVersion: kustomize.toolkit.fluxcd.io/v1
kind: Kustomization
metadata:
name: cluster-apps-local-path-provisioner
namespace: flux-system
labels:
substitution.flux.home.arpa/disabled: "true"
spec:
path: ./kubernetes/apps/kube-system/local-path-provisioner/app
prune: true
sourceRef:
kind: GitRepository
name: home-kubernetes
wait: false # no flux ks dependents
interval: 30m
retryInterval: 1m
timeout: 5m
@@ -0,0 +1,37 @@
---
apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
name: metrics-server
namespace: kube-system
spec:
interval: 30m
chart:
spec:
chart: metrics-server
version: 3.10.0
sourceRef:
kind: HelmRepository
name: metrics-server
namespace: flux-system
maxHistory: 2
install:
createNamespace: true
remediation:
retries: 3
upgrade:
cleanupOnFail: true
remediation:
retries: 3
uninstall:
keepHistory: false
values:
args:
- --kubelet-insecure-tls
- --kubelet-preferred-address-types=InternalIP,ExternalIP,Hostname
- --kubelet-use-node-status-port
- --metric-resolution=15s
metrics:
enabled: true
serviceMonitor:
enabled: true
@@ -0,0 +1,6 @@
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
namespace: kube-system
resources:
- ./helmrelease.yaml
Loaded 100 of 177 files, more files were not shown because too many files have changed in this diff. Show more