Merge 658802d580 into 7669fd4721

.gitignore

@@ -1,3 +1,6 @@
 .env/
 *.log
 ansible.cfg
+
+__pycache__/
+*.py[cod]

README.md

@@ -73,6 +73,21 @@ This requires at least k3s version `1.19.1` however the version is configurable
 
 If needed, you can also edit `inventory/my-cluster/group_vars/all.yml` to match your environment.
 
+#### verify routing prefixes and ip addresses in all.yml
+Ensure that the ip routing prefix matches the the one from the flannel_iface (`eth0` by default) for each host.
+Example `192.168.30` is the default routing prefix, and it is used in the following variables:
+- apiserver_endpoint
+- metal_lb_bgp_peer_address (commented by default)
+- metal_lb_ip_range
+Also Ensure that the apiserver_endpoint is not in the metal_lb_ip_range
+
+For your convience The playbook site.yml verifies the above elements of the config are valid.
+* Optionally to skip these verifications set, in all.yml `verify_config: false`  
+* Optionally to just verify the config Run
+    `ansible-playbook -i inventory/my-cluster/hosts.ini verify_config.yml`
+* Optionally manually recover ip routing prefix based on inet and netmask from
+    `ansible -i inventory/my-cluster/hosts.ini --become -m shell -a 'ifconfig eth0 | grep "inet "' all`
+
 ### ☸️ Create Cluster
 
 Start provisioning of the cluster using the following command:

inventory/sample/group_vars/all.yml

@@ -127,3 +127,5 @@ custom_registries_yaml: |
 #   HTTP_PROXY: "http://proxy.domain.local:3128"
 #   HTTPS_PROXY: "http://proxy.domain.local:3128"
 #   NO_PROXY: "*.domain.local,127.0.0.0/8,10.0.0.0/8,172.16.0.0/12,192.168.0.0/16"
+
+verify_config: true

roles/verify_config/tasks/main.yml

@@ -0,0 +1,13 @@
+---
+# To be ran on localhost after verify_config_gather role.
+# Verifies gatherd facts.
+- name: Collect all routing4_prefixes into a list
+  set_fact:
+    all_routing4_prefixes: "{{ groups['all'] | map('extract', hostvars, 'routing4_prefix') | list }}"
+    routing4_prefix: "{{ groups['all'] | map('extract', hostvars, 'routing4_prefix') | list | first }}"
+
+- name: Ensure all hosts have the same routing4_prefix
+  assert:
+    that: all_routing4_prefixes | unique | length == 1
+    fail_msg: "Not all hosts have the same routing4_prefix."
+    success_msg: "Using verified routing prefix {{ routing4_prefix }} across all hosts"

roles/verify_config_gather/filter_plugins/range_to_ips.py

@@ -0,0 +1,53 @@
+import netaddr
+
+''' returns True if ip is in range
+    examples:
+    - see test functions below
+'''
+def test(ip, range, expected):
+    assert netaddr_ip_in_dash_range(ip, range) == expected
+
+def test_netaddr_ip_in_dash_range():
+    # ipv4
+    test('192.168.1.1', '192.168.1.1-192.168.1.2', True)
+    test('192.168.1.1', '192.168.1.1', True)
+    test('192.168.1.1', '192.168.1.2-192.168.1.3', False)
+    test('192.168.1.1', '192.168.1.2', False)
+
+    # ipv6 style
+    test('::ffff:192.168.1.1', '::ffff:192.168.1.1-::ffff:192.168.1.8', True)
+    test('::ffff:192.168.1.1', '::ffff:192.168.1.1', True)
+    test('::ffff:192.168.1.1', '::ffff:192.168.1.2-::ffff:192.168.1.8', False)
+    test('::ffff:192.168.1.1', '::ffff:192.168.1.2', False)
+
+    # Note I expedted true but apperently the netaddr library does not support this?? or I don't understand ipv6 :)
+    test('::2:1', '::2:1-::2:2', False)
+    '''
+    todo: ?
+    - netaddr_ip_in_dash_range('192.168.1.1',  '192.168.1.0/24')             => True (TODO: test, implement)
+    - netaddr_ip_in_dash_range('192.168.99.1', '192.168.1.0/24')            => False (TODO: test, implement)
+    '''
+
+def netaddr_ip_in_dash_range(ip, range):
+    # return False early if range is invalid
+    if '-' not in range:
+        ip_start = range
+        ip_end = range
+    else:
+        ip_start = range.split('-')[0]
+        ip_end = range.split('-')[1]
+    return ip in [str(ip) for ip in netaddr.iter_iprange(ip_start, ip_end)]
+
+
+class FilterModule(object):
+        ''' Ansible filters. Interface to custom netaddr methods.
+            https://pypi.org/project/netaddr/
+        '''
+
+        def filters(self):
+            return {
+                'netaddr_ip_in_dash_range' : netaddr_ip_in_dash_range
+                }
+
+if __name__ == '__main__':
+    test_netaddr_ip_in_dash_range()

roles/verify_config_gather/tasks/main.yml

@@ -0,0 +1,117 @@
+---
+# To be ran on all hosts before role verify_config.
+# Gathers facts and ensures they are set
+
+- name: Set routing4_prefix from regex
+  set_fact:
+    routing4_prefix: "{{ hostvars[inventory_hostname]['ansible_' ~ flannel_iface]['ipv4']['broadcast']
+      | regex_replace('\\.?255', '') }}"
+    routing6_cidr: "{{ hostvars[inventory_hostname]['ansible_' ~ flannel_iface]['ipv6'][0]['address'] }}/{{
+      hostvars[inventory_hostname]['ansible_' ~ flannel_iface]['ipv6'][0]['prefix'] }}"
+
+- name: Check if fact routing4_prefix exists and is not empty
+  assert:
+    that:
+      - routing4_prefix is defined
+      - routing4_prefix is not none
+      - routing4_prefix != ''
+    fail_msg: >-
+      The fact 'routing_4prefix' is not defined, is null, or is empty
+      (based on flannel_iface: {{ flannel_iface }} ipv4 broadcast).
+
+# metal_lb_bgp_peer_address
+- name: Assert that metal_lb_bgp_peer_address starts with routing4_prefix, or is ipv6
+  assert:
+    that:
+      - >
+        metal_lb_bgp_peer_address.startswith(routing4_prefix)
+        or metal_lb_bgp_peer_address.matches('.*:.*')
+    fail_msg: >
+      The fact 'metal_lb_bgp_peer_address' <{{ metal_lb_bgp_peer_address }}>
+      doesn't start with the routing prefix <{{ routing4_prefix }}>
+  when: metal_lb_bgp_peer_address is defined
+
+# metal_lb_ip_range
+- name: >
+    Assert that metal_lb_ip_range (when string) contains <ipv4>-<ipv4>
+    and both ips start with routing4_prefix, skip any containing ':' (ipv6)
+  assert:
+    that:
+      - metal_lb_ip_range | regex_search('^{{ routing4_prefix }}\.[0-9]{1,3}-{{ routing4_prefix }}\.[0-9]{1,3}$|.*:.*')
+    fail_msg: >
+      metal_lb_ip_range <{{ metal_lb_ip_range }}> has one or more ipv4s
+      that don't start with the routing prefix <{{ routing4_prefix }}>
+  when: metal_lb_ip_range is string
+
+- name: Assert that metal_lb_ip_ranges (when list) has only strings that match the regexes in the task above
+  assert:
+    that:
+      - >
+        ( metal_lb_ip_range
+        | select('match', '^{{ routing4_prefix }}\.[0-9]{1,3}-{{ routing4_prefix }}\.[0-9]{1,3}$|.*:.*')
+        | list | length
+        )
+        ==
+        (metal_lb_ip_range
+        | length
+        )
+    fail_msg: >
+      metal_lb_ip_range <{{ metal_lb_ip_range }}> has one or more values with ipv4s
+      that don't start with the routing prefix <{{ routing4_prefix }}>
+  when: metal_lb_ip_range is not string and metal_lb_ip_range is not mapping and metal_lb_ip_range is iterable
+
+# apiserver_endpoint
+- name: Assert that apiserver_endpoint is not in metal_lb_ip_range (when string) using network_in_usable
+  # For /<mask> ranges
+  assert:
+    that:
+      - not ( metal_lb_ip_range | ansible.utils.network_in_usable( apiserver_endpoint ))
+    fail_msg: "apiserver_endpoint {{ apiserver_endpoint }} cannot be in the metal_lb_ip_range {{ metal_lb_ip_range }}"
+    success_msg: >
+      apiserver_endpoint {{ apiserver_endpoint }} is *probably* not in the metal_lb_ip_range {{ metal_lb_ip_range }}
+  when: metal_lb_ip_range is string
+
+- name: >
+    Assert that apiserver_endpoint is not in metal_lb_ip_range (when string)
+    using custom filter netaddr_ip_in_dash_range
+  # For <ip>-<ip> ranges. Not sure this works for ipv6
+  assert:
+    that:
+      - not (apiserver_endpoint | netaddr_ip_in_dash_range(metal_lb_ip_range))
+    fail_msg: "apiserver_endpoint {{ apiserver_endpoint }} cannot be in the metal_lb_ip_range {{ metal_lb_ip_range }}"
+    success_msg: >
+      apiserver_endpoint {{ apiserver_endpoint }} is *probably* not in the metal_lb_ip_range {{ metal_lb_ip_range }}
+  when: metal_lb_ip_range is string
+  # *probably* in the success_msg sections of the prior two tasks because not all cases may work
+
+- name: Assert that apiserver_endpoint is not in metal_lb_ip_ranges (when list) using logic of the task above
+  assert:
+    that:
+      - not (apiserver_endpoint | netaddr_ip_in_dash_range(item))
+    # this probably fails on an ipv6 range ''::1-::2', and on <ipv4or6>/<mask> ranges
+    fail_msg: "apiserver_endpoint {{ apiserver_endpoint }} cannot be in the metal_lb_ip_range item {{ item }}"
+    success_msg: >
+      apiserver_endpoint {{ apiserver_endpoint }} is *probably* not in the metal_lb_ip_range item {{ item }}
+  loop: "{{ metal_lb_ip_range | list }}"
+  when: metal_lb_ip_range is not string   and metal_lb_ip_range is not mapping   and metal_lb_ip_range is iterable
+  # these (when string, when list tasks) smell funny.
+  # It seems there should be a way in one task to handle an object that is a string or a list
+  #    and loop on {{ [metal_lb_ip_range] }}  or {{ metal_lb_ip_range }} respectively
+  # when it is a string it skips each char :(
+  # I tried the select pattern like in the task
+  #    'assert that metal_lb_ip_ranges (when list) has only strings that match the regexes in the task above'
+  #  but it didn't work
+
+- name: Assert that apiserver_endpoint, is ipv4 and starts with routing4_prefix, or is ipv6 and is in routing6_cidr
+  assert:
+    that:
+      - apiserver_endpoint is defined
+      - apiserver_endpoint is not none
+      - >-
+        ( apiserver_endpoint | ansible.utils.ipv4 and apiserver_endpoint.startswith(routing4_prefix) )
+        or
+        ( apiserver_endpoint | ansible.utils.ipv6 and apiserver_endpoint | ansible.utils.ipaddr(routing6_cidr))
+    fail_msg: >
+      The fact 'apiserver_endpoint' <{{ apiserver_endpoint }}>
+      doesn't start with the routing prefix <{{ routing4_prefix }}>
+      or is not in the routing6_cidr <{{ routing6_cidr }}>

site.yml

@@ -1,4 +1,7 @@
 ---
+- name: Import verify_config playbook
+  ansible.builtin.import_playbook: verify_config.yml
+
 - name: Prepare Proxmox cluster
   hosts: proxmox
   gather_facts: true

verify_config.yml

@@ -0,0 +1,14 @@
+---
+- name: Gather config for verify
+  hosts: all
+  gather_facts: true
+  roles:
+    - role: verify_config_gather
+      when: verify_config is not defined or verify_config
+
+- name: Verify config
+  hosts: localhost
+  gather_facts: false
+  roles:
+    - role: verify_config
+      when: verify_config is not defined or verify_config