Manage multiple Linux servers: tools that work
SSH config, tmux, Ansible, Uptime Kuma, Zabbix and Webmin, ranked by how many servers you have: what each replaces, setup cost in minutes, the one gotcha.
#ansible
Filtering by topic #ansible · clear
Ansible playbook vs role: when to use each
When a flat Ansible playbook is enough and when a role earns its directories: the role layout, ansible-galaxy init, role calls, and variable precedence.
Ansible playbook to install Zabbix
An idempotent Ansible playbook that installs Zabbix server on Ubuntu 24.04, a vaulted database password that actually wins, and agents across your inventory.
Ansible check mode and --diff dry runs
Ansible check mode explained: what --check and --diff really prove, and where a dry run gives you a wrong answer before you apply for real.
Run one Ansible playbook from another
import_playbook, import_tasks and include_tasks do different jobs. See which one your site.yml needs, and prove it with ansible-playbook --list-tasks.
Ansible inventory files explained
Your Ansible inventory decides which host gets which variable: INI and YAML forms, groups, group_vars, host_vars, connection settings and precedence.
Ansible Vault: encrypt secrets in git
Keep passwords and API tokens in a playbook repo safely: encrypt a vars file or one inline string, split staging from production, and rekey cleanly.
Ansible vs Terraform: which do you need?
Terraform creates the VPS, Ansible configures it. The real split, why provisioners fight both tools, the handoff commands, and when you need only Ansible.
Ansible tutorial: your first playbook on a VPS
Install Ansible with pipx on Ubuntu 24.04, write an inventory and a first playbook that hardens a fresh VPS, plus fixes for Permission denied and sudo errors.
Ansible: ignore unreachable hosts
Ansible treats an unreachable host differently from a failed task. Use ignore_unreachable, serial and max_fail_percentage, and still see what was missed.
Ansible facts: when to turn gather_facts off
Learn what Ansible facts cost, when gather_facts: false is the right call, and how gather_subset and JSON fact caching cut the wait on every run.
Ansible reboot: only when the server needs it
Gate the reboot on evidence the host reports: the Ubuntu marker file, the Red Hat exit code, then reboot in slices so the whole fleet stays up.
Ansible templates and handlers, by example
Render an nginx config from a Jinja2 template, notify a handler so it reloads only on a real change, then run the play twice to prove idempotence.