Ansible Collection - adfinis.maintenance

This Ansible collection provides a framework to run maintenance tasks automatically against a host.

It does so in a way that may be a bit counterintuitive for someone used to Ansible:

• Almost everything is executed in check mode - The exceptions are tasks that won't break anything.
• Tasks that report ok can be considered completed - no human intervention is required.
• Tasks that report changed or failed require human intervention.
  • NOTE: It is expected that tasks fail if some critical prerequesite is not met. Usually this is not a bug, and the correct way to resolve the issue is to either fulfil the required prerequisite, or to exclude the failing task (see maintenance_exclude_tasks below).
• The plugins in this collection are designed for Python 3. If you have a system where the default python interpreter is still Python 2, you need to add ansible_python_interpreter=/usr/bin/python3 to the hosts' hostvars.

The tasks to run are grouped into "checklists". By convention, each checklist is implemented in a separate role, e.g.:

• The maintenance_10_linux role implements tasks common to most Linux systems
• The maintenance_11_debian role implements Debian-specific tasks
• The maintenance_14_sles role implements Suse-specific tasks

Checklists (i.e. roles) are assigned to hosts via playbooks. The example playbook in playbooks/playbook.yml applies each role to an eponymous host group. When using this example playbook, your inventory could look like this:

[maintenance_10_linux:children]
maintenance_11_debian
maintenance_15_rhel

[maintenance_11_debian]
debian01.example.org
debian02.example.org

[maintenance_14_sles]
sles01.example.org
sles02.example.org

[maintenance_15_rhel]
rhel01.example.org
centos01.example.org

Every checklist task (which may consist of more than one Ansible task) has a unique task ID in the form XX-YYY which XX being the checklist id and YYY being a consecutive number within the checklist. E.g. 10-042. These IDs can be used to exclude tasks from being applied to one or more hosts using the maintenance_exclude_tasks hostvar. To be able to easily exclude some checks globally in a playbook and some checks only on a host-level, the variable by default combines maintenance_global_exclude_tasks and maintenance_host_exclude_tasks:

# Exclude tasks globally
--- group_vars/all/maint.yml
maintenance_global_exclude_tasks:
  - 10-042

--- host_vars/my_host/maint.yml
# Additionally exclude 11-023 only on one host
maintenance_host_exclude_tasks:
  - 11-023
  # 10-042 is implicitly excluded by above global statement

--- host_vars/other_host/maint.yml
# Explicitly ONLY exclude a task on my host
maintenance_exclude_tasks:
  - 11-023
  # 10-042 will be included, global list doesn't apply

Some of the checklists have additional options which can be overwritten at the host or group level, e.g. to override the expected alias for root in /etc/aliases:

linux_serverlogs_root_alias: [email protected]

Check out roles/maintenance_*/defaults/main.yml to see which options can be overwritten.

Recommendations for ansible.cfg

[defaults]
display_skipped_hosts=no
display_ok_hosts=no
callback_whitelist=adfinis.maintenance.report
callbacks_enabled=adfinis.maintenance.report
duplicate_dict_key=ignore
inject_facts_as_vars=no
collections_path=./galaxy
roles_path=./galaxy/roles

Example output

Running the example playbook in playbooks/playbook.yml with the recommended settings above against a host that is in the maintenance_10_linux and maintenance_11_debian hostgroups will provide an output like this:

user@maintenancemaster:~/git/maintenance-test$ ansible-playbook -l debian01.example.org playbook.yml 

PLAY [Run automated maintenance tasks] *****************************************************************************************************************************************************************************************************************************************************

TASK [adfinis.maintenance.maintenance_10_linux : 10-041: Security: Logins: Are there logins from suspicious hosts/users? | Gather past logins from `last` output] **************************************************************************************************************************
changed: [debian01.example.org]

TASK [adfinis.maintenance.maintenance_10_linux : 10-042: Security: SSH keys: Check for unknown or outdated keys for root and all users] ****************************************************************************************************************************************************
changed: [debian01.example.org]

TASK [adfinis.maintenance.maintenance_11_debian : 11-016: dpkg status: Are there packages which do not have the dpkg status ii or hi? | Report matching packages] **************************************************************************************************************************
changed: [debian01.example.org] => {
    "debian_dpkg_status.stdout_lines": [
        "rc  linux-image-5.10.0-9-amd64           5.10.70-1                      amd64        Linux 5.10 for 64-bit PCs (signed)"
    ]
}

TASK [adfinis.maintenance.maintenance_11_debian : 11-017: apt: Simulate the package upgrade] ***********************************************************************************************************************************************************************************************
changed: [debian01.example.org]

PLAY RECAP *********************************************************************************************************************************************************************************************************************************************************************************
debian01.example.org : ok=35   changed=4    unreachable=0    failed=0    skipped=0    rescued=0    ignored=0   


debian01.example.org
- [x] 10-028: Systemd: Check all important service units
- [x] 10-032: Disk: Fstab: fstab correct?
- [x] 10-034: Disk: logrotate: Are there files in /var/log that are not rotated?
- [x] 10-035: Are there logfiles outside /var/log that are not rotated?
- [x] 10-039: Logfiles: Does journald log to persistent storage?
- [x] 10-040: Security: User: Are all created users documented in the wiki with password?
- [ ] 10-041: Security: Logins: Are there logins from suspicious hosts/users?
- [ ] 10-042: Security: SSH keys: Check for unknown or outdated keys for root and all users
- [~] 10-050: Mail: serverlogs: Is serverlogs@ entered in /etc/aliases for root?
- [~] 10-051: Mail: aliases.db: Make sure /etc/aliases.db is up to date
- [x] 11-011: Security: Are the security updates in the sources.list?
- [x] 11-012: Repository: Check if repository is set to release name (e.g. 'bullseye') and not to 'stable'
- [x] 11-013: For old distributions, has the repository been moved to http://archive.debian.org/ already?
- [x] 11-014: Update package lists and check for errors
- [ ] 11-016: dpkg status: Are there packages which do not have the dpkg status ii or hi?
- [ ] 11-017: apt: Simulate the package upgrade
- [x] 11-019: apt: Remove obsolete packages
- [x] 11-020: boot-config: Check boot configuration: keep bootloader up to date
- [x] 10-061: Updates: Check if a major update is pending.

With the ansible.cfg settings above, tasks that completed with the status ok or skipped won't be displayed, instead only changed tasks are shown, getting you a clear report on that you should look at.

The content of the checkboxes has the following meaning:

- [x] ok - nothing to do
- [~] skipped - task skipped as configured. Check if it's still reasonable to skip this task
- [ ] changed or failed - See output on what to do

If you want more detailed output, execute the playbook with the --diff option, example output:

TASK [adfinis.maintenance.maintenance_10_linux : 10-042: Security: SSH keys: Check for unknown or outdated keys for root and all users] ****************************************************************************************************************************************************
--- before: authorized_keys (root)
+++ after: authorized_keys (root)
@@ -1 +0,0 @@
-ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABgQCoquOVPUnXKNP25SQzdKpXKby2s1fDhZS/zllKW5zGMr+C9mnf7xMN+sB16yfXhQRCJGWjzjxNPl56lB9s4jV1lrFtDVEmGu+arv2eQa1cQJ6ggeOxhzfpbPVJh0T5cZg9XpuucJDTFceA/wN5eeWAIAQpzjeFTYn0obDjzSzoXsPiRZ35URCEF6R1/+6gj6WaosiGiCVUyyIK5vJLsbJCVsV+hSFmTrZfKIt33h+XcjKacfzGNsON++2B5m0EEvCy0= user@maintenancemaster

changed: [debian01.example.org]

There is also a checklist summarising all tasks that were run but finished with either ok or skipped.

Development Setup

For development on this collection, we recommend the following setup:

• Set up a separate project folder with an inventory of your development target systems, hostvars and ansible.cfg as described above.
• cd to that folder.
• Run ansible-galaxy collection install adfinis.maintenance. This will also install all the required dependencies.
• rm -rf ./galaxy/ansible_collections/adfinis/maintenance to remove the collection downloaded from Galaxy.
• git clone github.com:adfinis/ansible-collection-maintenance ./galaxy/ansible_collections/adfinis/maintenance to checkout the git repository.
• Do your development inside ./galaxy/ansible_collections/adfinis/maintenance.
• For testing, run the playbook from your project folder as e.g. ansible-playbook -i inventory --diff adfinis.maintenance.playbook.
  • If your Ansible version does not allow this yet, use the playbook path rather than its FQCN, i.e. ./galaxy/ansible_collections/adfinis/maintenance/playbooks/playbook.yml.
  • To only test a single checklist item, you can run ansible-playbook limited to the taskid, using the maintenance_only variable: ansible-playbook -e maintenance_only=10-011 ....

License

GPL-3.0

Author Information

adfinis.maintenance was written by:

• Adfinis AG | Website | Twitter | GitHub