The best configuration management tools in 2026
Ansible is the best configuration management tool for most Linux teams: agentless over SSH, readable YAML and free (GPL-3.0). Salt is the best for large fleets that need fast remote execution, and it's still Apache-2.0 and actively released under Broadcom. Puppet is the best for enterprises that want a supported agent that corrects drift every 30 minutes, though its builds are now proprietary.
Looking for a SaltStack alternative? Pick Ansible if you want no agent, OpenVox (the free Puppet fork) if you want a pull-based agent, or pyinfra if you'd rather write Python. Terraform and OpenTofu provision infrastructure; they don't replace these tools.
Our top picks
- AnsibleBest for most Linux teams: agentless automation over SSH
- SaltBest for large fleets and event-driven automation
- PuppetBest for enterprises that want a supported, drift-correcting agent
Compare at a glance
| Tool | Best for | Pricing | Open source | Deployment | Tested by us |
|---|---|---|---|---|---|
| Ansible | most Linux teams: agentless automation over SSH | Free (open source); Red Hat Ansible Automation Platform: no public price (contact sales) | Yes | Self-hosted CLI, agentless over SSH (Red Hat Ansible Automation Platform is the commercial edition) | Yes |
| Salt | large fleets and event-driven automation | Free (open source); the commercial VMware edition is a licensed add-on with no public price | Yes | Self-hosted (master and minion agents, or agentless Salt SSH) | Yes |
| Puppet | enterprises that want a supported, drift-correcting agent | Puppet Core Developer license free for fewer than 25 nodes; Puppet Core Commercial and Puppet Enterprise: custom pricing (contact sales) | No | Self-hosted (agent plus primary server) | Docs only |
| OpenVox | Puppet users who want it free and open source | Free (open source); support only from third-party companies | Yes | Self-hosted (agent plus server) | Yes |
| Chef Infra | Ruby-friendly teams that want a supported SaaS | SaaS Business $59 and Enterprise $189 per node per year; Enterprise Plus and self-managed: quote; free license for personal, non-commercial use only | No | SaaS or self-managed (Chef Infra Server plus client; agentless mode for Linux) | Docs only |
| pyinfra | Ansible-style automation in plain Python | Free (open source) | Yes | Self-hosted CLI, agentless over SSH | Yes |
| CFEngine | lightweight agents on many hosts | Community: free (open source); Enterprise: free for up to 25 hosts without support, then by quote | Yes | Self-hosted (agent on every host, policy hub) | Docs only |
| Rudder | compliance-focused teams that want a web interface | Rudder Core: free (open source); Enterprise: per node per year, no public price (contact sales) | Yes | Self-hosted (server, relays and agents) | Docs only |
| Cinc | free, EULA-free Chef binaries | Free (open source) | Yes | Self-hosted (drop-in replacement for chef-client) | Docs only |
| mgmt | experimenting with real-time, event-driven configuration | Free (open source); enterprise support offered by a partner with no public price | Yes | Self-hosted (Go binary; agents can form a cluster) | Docs only |
The full ranking
Ansible
Best for: most Linux teams: agentless automation over SSH
Ansible is the default choice for Linux configuration management for good reasons: there's no agent to install, it connects over SSH, and playbooks are YAML that most admins can read on day one. Managed hosts need only SSH and Python, and Windows works over WinRM, PSRP or SSH. It's GPL-3.0 and very active: ansible-core 2.22.0 shipped on 2 November 2026. In our test, ad-hoc runs reached a Debian 12.15 and an AlmaLinux 9.8 server with nothing installed on either. The trade-off is the push model: nothing re-applies your configuration between runs unless you schedule them.
- Pricing
- Free (open source); Red Hat Ansible Automation Platform: no public price (contact sales) (checked November 15, 2026, source)
- License
- Open source (GPL-3.0-or-later)
- Deployment
- Self-hosted CLI, agentless over SSH (Red Hat Ansible Automation Platform is the commercial edition)
- Runs on
- Control node on Linux, macOS or WSL; manages any host with SSH and Python, plus Windows over WinRM, PSRP or SSH
Pros
- No agent: SSH and Python on each server are enough
- Readable YAML playbooks
- Free (GPL-3.0) with frequent releases
Cons
- Push model: state isn't enforced between runs
- Managed hosts need Python
- Commercial platform price is quote-only
Tested by us: ansible-core 2.22.0 from a workstation: fact gathering and check-mode apt and dnf runs on Debian 12.15 and AlmaLinux 9.8, 14 Nov 2026 (UTC); nothing installed on the servers
Salt
Best for: large fleets and event-driven automation
Salt pairs configuration management with fast remote execution: minions stay connected to a master over ZeroMQ, and states are YAML with Jinja. It also runs masterless, and Salt SSH works with no agent at all, though Salt's own docs say it's substantially slower. Under Broadcom it's still Apache-2.0 and actively released: 3008.5 LTS, 3006.30 LTS and 3007.17 all came out on 26 October 2026, with packages now on packages.broadcom.com. In our test, salt-minion 3008.5 installed from that repository on AlmaLinux 9.8, left its service disabled, and ran states masterless with salt-call --local.
- Pricing
- Free (open source); the commercial VMware edition is a licensed add-on with no public price (checked November 15, 2026, source)
- License
- Open source (Apache-2.0)
- Deployment
- Self-hosted (master and minion agents, or agentless Salt SSH)
- Runs on
- Master and minion on RHEL, AlmaLinux, Rocky and Oracle Linux, Debian 11–13, Ubuntu 22.04–24.04, SLES, Fedora, Photon OS, Amazon Linux; minions on Windows and macOS
Pros
- Fast remote execution plus states
- Agent, masterless or agentless (Salt SSH)
- Apache-2.0, with one LTS release a year
Cons
- A master and minions to run and secure
- Salt SSH is much slower than the agent, per its docs
- Commercial edition is tied to VMware Cloud Foundation
Tested by us: salt-minion 3008.5 from the Salt Project's repository on AlmaLinux 9.8, masterless salt-call --local (grains, pkg.list_upgrades, a file state with test=True), 14 Nov 2026 (UTC); removed afterwards with its repository, signing key and user
Puppet
Best for: enterprises that want a supported, drift-correcting agent
Puppet is the classic model-driven agent: each server pulls a catalog compiled from Puppet-language manifests, every 30 minutes by default, and corrects any drift. It's mature, with agents for current RHEL-family, Debian, Ubuntu, SLES, Windows and macOS releases. Its licensing has changed, though: Perforce moved new builds to a private location in 2025, and its docs say that since February 2026 it develops Puppet Core 8 and 9 as a proprietary product. A Developer license is free for fewer than 25 nodes; above that, pricing is custom. Teams that want Puppet without the license should look at OpenVox.
- Pricing
- Puppet Core Developer license free for fewer than 25 nodes; Puppet Core Commercial and Puppet Enterprise: custom pricing (contact sales) (checked November 15, 2026, source)
- License
- Proprietary (Puppet Core and Enterprise); older open-source Puppet releases are Apache-2.0
- Deployment
- Self-hosted (agent plus primary server)
- Runs on
- Agents for RHEL, AlmaLinux and Rocky 8–10, Debian 11–13, Ubuntu 22.04–26.04, SLES 12 and 15, Amazon Linux 2023, Fedora, Windows and macOS; primary server on Linux
Pros
- Continuous drift correction by default
- Mature, with wide platform support
- Vendor support, CVE SLAs and compliance modules
Cons
- Proprietary builds since 2025
- Free only below 25 nodes
- Custom pricing, no public list
Not tested by us yet: based on the vendor's documentation.
OpenVox
Best for: Puppet users who want it free and open source
OpenVox is the community's answer to Puppet going proprietary: a soft fork from Vox Pupuli that keeps publishing Apache-2.0 agent, server and database packages compatible with Puppet code. It's current, with openvox 9.4.0 and 8.33.0 released on 13 November 2026 and packages for EL 8–10, Debian 12–13, Ubuntu 22.04–26.04, SLES, Fedora and Amazon Linux. In our test, openvox-agent 9.4.0 installed from Vox Pupuli's apt repository on Debian 12.15 and compiled and dry-ran a three-resource manifest in under 3 seconds. Support comes only from third-party companies, and it can't be installed alongside Puppet.
- Pricing
- Free (open source); support only from third-party companies (checked November 15, 2026, source)
- License
- Open source (Apache-2.0)
- Deployment
- Self-hosted (agent plus server)
- Runs on
- Packages for EL 8–10, Fedora, Amazon Linux 2023, SLES 15–16, Debian 12–13, Ubuntu 22.04–26.04, plus Windows and macOS agents
Pros
- Free, Apache-2.0 and compatible with Puppet code
- Active releases
- Packages for every major distribution
Cons
- Community project: support only from third parties
- Can't share a host with Puppet packages
- Same learning curve as Puppet
Tested by us: openvox-agent 9.4.0 from apt.voxpupuli.org on Debian 12.15: puppet apply --noop of a file, package and service manifest, 14 Nov 2026 (UTC); purged afterwards with its repository
Chef Infra
Best for: Ruby-friendly teams that want a supported SaaS
Chef Infra applies cookbooks written in a Ruby DSL, usually pulled from a Chef Infra Server, and Progress now sells it inside Chef 360 with public SaaS prices: $59 per node per year for Business and $189 for Enterprise. The source is Apache-2.0, but Progress's binaries come under the Chef EULA, and the free license covers only personal and non-commercial use. Chef Infra Client 19 supports current RHEL, Debian, Ubuntu, SLES and Amazon Linux releases plus Windows, macOS and FreeBSD, and an agentless mode runs over SSH on Linux. For free production use, Cinc builds the same code without the EULA.
- Pricing
- SaaS Business $59 and Enterprise $189 per node per year; Enterprise Plus and self-managed: quote; free license for personal, non-commercial use only (checked November 15, 2026, source)
- License
- Apache-2.0 source; Progress's binaries are under the Chef EULA (free license for personal, non-commercial use)
- Deployment
- SaaS or self-managed (Chef Infra Server plus client; agentless mode for Linux)
- Runs on
- Chef Infra Client 19 on RHEL 8–10, Oracle Linux, Amazon Linux 2023, Debian 12–13, Ubuntu 22.04–26.04, SLES 15–16, Windows, macOS and FreeBSD
Pros
- Public per-node SaaS prices
- Agent or agentless mode on Linux
- Ruby DSL handles complex logic
Cons
- Free license is non-commercial only
- Ruby DSL is a steeper start than YAML
- Self-managed pricing is quote-only
Not tested by us yet: based on the vendor's documentation.
pyinfra
Best for: Ansible-style automation in plain Python
pyinfra is the best Ansible alternative for people who'd rather write Python than YAML: your code becomes shell commands that pyinfra runs over SSH, locally or in Docker containers. Targets need only shell access. It's MIT-licensed and active, with 3.11.1 released on 12 November 2026. In our test it gathered distribution facts from a Debian and an AlmaLinux server in under 5 seconds and left nothing behind on either. It's a smaller project with no commercial support, its tested-target list lags current releases, and version 3 dropped Windows targets.
- Pricing
- Free (open source) (checked November 15, 2026, source)
- License
- Open source (MIT)
- Deployment
- Self-hosted CLI, agentless over SSH
- Runs on
- Controller on Linux, macOS or Windows; targets need only shell access (Linux distributions, BSDs, Docker containers); no Windows targets in v3
Pros
- Plain Python instead of YAML
- Agentless: targets need only shell access
- Fast, and leaves nothing on the server
Cons
- No commercial support
- No Windows targets in version 3
- Tested-target list lags current distro releases
Tested by us: pyinfra 3.11.1 from a workstation: fact server.LinuxDistribution on Debian 12.15 and AlmaLinux 9.8, plus a --dry server.packages run, 14 Nov 2026 (UTC); nothing installed on the servers
CFEngine
Best for: lightweight agents on many hosts
CFEngine runs a lightweight C agent on every host that pulls promise-based policy from a hub, by default every 5 minutes, and keeps enforcing it if the hub is down. The Community edition is GPL-3.0, and CFEngine's comparison page says you can try CFEngine Enterprise for free on up to 25 hosts; beyond that it's priced by quote. Releases are steady: 3.29.0 came out on 6 November 2026, 3.27 is the long-term support line, and 3.29 can run Ansible playbooks from CFEngine. Its policy language is its own and takes time to learn, and 3.27 dropped Solaris, AIX, HP-UX, SUSE and Debian 11.
- Pricing
- Community: free (open source); Enterprise: free for up to 25 hosts without support, then by quote (checked November 15, 2026, source)
- License
- Open core (Community GPL-3.0; Enterprise commercial)
- Deployment
- Self-hosted (agent on every host, policy hub)
- Runs on
- Hub on RHEL 8–10, Debian 12–13, Ubuntu 22.04–24.04; agents on those plus RHEL 7, Ubuntu 20.04 and Windows Server
Pros
- Lightweight agent with frequent runs
- Keeps enforcing policy if the hub is down
- Enterprise free to try on up to 25 hosts
Cons
- Its own policy language to learn
- 3.27 dropped SUSE, Debian 11 and commercial Unix
- Enterprise price above 25 hosts by quote
Not tested by us yet: based on the vendor's documentation.
Rudder
Best for: compliance-focused teams that want a web interface
Rudder is configuration management built around compliance: an agent on every node applies and audits its rules, and a web interface shows how each node complies, which suits security baselines and audits. Rudder Core is GPL-3.0, but the free edition has firm limits: free packages cover only current distribution releases (Debian 12–13, Ubuntu 24.04–26.04, RHEL-family 9–10, SLES 15 SP6+, Amazon Linux 2023). Older distros, Windows and feature plugins such as patch management need a subscription, priced per node per year by quote. Releases are current, with 9.2.1 out on 4 November 2026.
- Pricing
- Rudder Core: free (open source); Enterprise: per node per year, no public price (contact sales) (checked November 15, 2026, source)
- License
- Open source core (GPL-3.0); feature plugins need a subscription
- Deployment
- Self-hosted (server, relays and agents)
- Runs on
- Free packages for the latest Debian, Ubuntu, RHEL-family, SLES and Amazon Linux releases; older distros and Windows need a subscription
Pros
- Web interface with compliance reporting
- Free core for current distributions
- Active releases
Cons
- Older distros and Windows need a subscription
- Patch management is a paid plugin
- No public price
Not tested by us yet: based on the vendor's documentation.
Cinc
Best for: free, EULA-free Chef binaries
Cinc is a community rebuild of Chef's open-source code, changed only in branding, so you can run Chef Infra in production without the Progress EULA. Cinc Client is a drop-in replacement for chef-client, and it's current: 19.4.46 shipped on 11 November 2026 for AlmaLinux, Rocky, Debian, Ubuntu, SLES, Amazon Linux, FreeBSD, macOS and Windows. The project says plainly that it comes with no warranty or support, and it says Progress is retiring Chef Infra Server in favor of Chef 360, so Cinc Server is becoming an independent fork. Pick it if you already know Chef and need it free.
- Pricing
- Free (open source) (checked November 15, 2026, source)
- License
- Open source (Apache-2.0), no EULA
- Deployment
- Self-hosted (drop-in replacement for chef-client)
- Runs on
- Cinc Client 19 on AlmaLinux and Rocky 8–10, Amazon Linux 2023, Debian 12–13, Ubuntu 22.04–26.04, SLES 15–16, FreeBSD, macOS and Windows
Pros
- Free for commercial use (Apache-2.0, no EULA)
- Drop-in replacement for chef-client
- Keeps pace with upstream Chef
Cons
- No warranty or support
- Server side is becoming an independent fork
- Same Ruby learning curve as Chef
Not tested by us yet: based on the vendor's documentation.
mgmt
Best for: experimenting with real-time, event-driven configuration
mgmt is a real and active project: a Go engine that reacts to events as they happen instead of running on a schedule, with its own small language, mcl, and agents that can form a cluster with embedded etcd. It's GPL-3.0, released 1.1.0 in June 2026, and had commits as recently as 14 November 2026. Its README says it's used in production, but there's no versioned distribution support list, and its FAQ says it works best on Linux. Try it for new ideas; don't move a fleet onto it yet.
- Pricing
- Free (open source); enterprise support offered by a partner with no public price (checked November 15, 2026, source)
- License
- Open source (GPL-3.0)
- Deployment
- Self-hosted (Go binary; agents can form a cluster)
- Runs on
- Linux-focused; no versioned distribution list
Pros
- Event-driven, real-time changes
- Built-in clustering with embedded etcd
- Free (GPL-3.0)
Cons
- Its own language (mcl) to learn
- No versioned support list
- Enterprise support price isn't public
Not tested by us yet: based on the vendor's documentation.
How we picked
We compared the configuration management tools that Linux teams actually choose between, plus the forks that appeared after recent licensing changes. Each was ranked on:
- Fit for Linux servers: supported distributions and how much has to be installed on each host.
- Day-to-day effort: the language you write in, and agent or agentless.
- License and cost: the license of the code and the binaries you'd actually run, and prices from vendor pages checked on 15 Nov 2026.
- Health: recent releases and who maintains it.
We ran four of them on our test servers on 14 Nov 2026 (UTC): Ansible and pyinfra from a workstation against a Debian 12.15 and an AlmaLinux 9.8 VM, Salt masterless on the AlmaLinux VM, and OpenVox on the Debian VM. We removed everything afterwards, including package repositories, signing keys and service users. The rest come from their documentation. We don't make a configuration management tool.
What we saw on our test servers
Ansible and pyinfra needed nothing on the servers. pyinfra read both servers' distribution details in under 5 seconds:
$ pyinfra web-01,web-02 --ssh-user root -y fact server.LinuxDistribution
...
"name": "AlmaLinux",
"major": 9,
"minor": 8,
Salt 3008.5 installed from the Salt Project's own repository (now on packages.broadcom.com), with the salt-minion service left disabled, which is what you want for masterless use. A state run in test mode showed what it would change:
$ salt-call --local --file-root=/root/salt-test state.apply motd test=True
...
Summary for local
------------
Succeeded: 1 (unchanged=1, changed=1)
Failed: 0
OpenVox 9.4.0 installed from Vox Pupuli's apt repository, and puppet apply --noop compiled a three-resource manifest (a file, a package and a service) in 0.22 seconds:
Notice: Compiled catalog for web-01 in environment production in 0.22 seconds
Notice: /Stage[main]/Main/File[/tmp/openvox-test-banner.txt]/ensure: defined content as '{sha256}82eb5260...' (noop)
Notice: Applied catalog in 0.11 seconds
Note that installing openvox-agent on Debian enabled its puppet service without starting it; disable it if you only use puppet apply.
Looking for a SaltStack alternative?
First, check why you're leaving. If it's fear that Salt is dying, the evidence says otherwise: the Salt Project, now part of Broadcom, shipped 3008.5 LTS, 3006.30 LTS and 3007.17 on 26 October 2026, publishes one LTS a year, and the code is still Apache-2.0. What did change is where packages live (packages.broadcom.com) and the commercial edition, which is now tied to VMware Cloud Foundation. If you still want out, the right alternative depends on what you used Salt for. For configuration plus ad-hoc commands across servers, Ansible is the closest everyday replacement and needs no master or minions. For agents that keep enforcing state, OpenVox or Puppet do that with a pull model every 30 minutes, and CFEngine every 5. For writing logic in Python rather than YAML and Jinja, pyinfra. If you depend on Salt's event bus and reactors, nothing here is a drop-in replacement; mgmt is the closest in spirit but far less mature.
What are the best Ansible alternatives?
People usually leave Ansible for one of three reasons, and each has a different answer. If YAML and Jinja templating got painful, pyinfra keeps the same agentless SSH model but lets you write plain Python, and Chef or Cinc give you a full Ruby DSL. If push runs are the problem, because servers drift between runs or because large fleets take too long over SSH, an agent-based tool fixes that: Salt for fast, event-driven execution, Puppet or OpenVox for scheduled drift correction, CFEngine for a very small agent on many hosts. If you need a central web interface with compliance reports, Rudder is built for that, and Red Hat sells Ansible Automation Platform as the supported edition of Ansible. Before you switch, remember what Ansible does well: no agent on any server and the easiest start of any tool here. Many teams end up using Ansible for one-off changes alongside an agent-based tool.
Agent or agentless, push or pull?
| Tool | On each server | Model | You write | License |
|---|---|---|---|---|
| Ansible | Nothing (SSH and Python) | Push | YAML | GPL-3.0 |
| pyinfra | Nothing (shell access) | Push | Python | MIT |
| Salt | Minion, or nothing with Salt SSH | Push and event-driven; masterless possible | YAML and Jinja | Apache-2.0 |
| Puppet / OpenVox | Agent | Pull, every 30 minutes by default | Puppet language | Proprietary / Apache-2.0 |
| Chef / Cinc | Client (agentless mode on Linux) | Pull | Ruby DSL | EULA binaries / Apache-2.0 |
| CFEngine | Agent | Pull, every 5 minutes by default | CFEngine policy language | GPL-3.0 core |
| Rudder | Agent | Rules and compliance reports | Rules in a web interface | GPL-3.0 core |
| mgmt | Agent (optional cluster) | Event-driven, real time | mcl | GPL-3.0 |
Agentless tools are easier to start with and leave nothing running on your servers. Agent-based tools keep servers in line between your runs and scale better to large fleets, at the cost of one more service to patch on every host.
Configuration management or infrastructure as code?
They overlap in name, not in job. Terraform and OpenTofu create and change infrastructure through provider APIs: virtual machines, networks, load balancers, DNS records. Configuration management tools work inside each machine: packages, files, users, services. HashiCorp's own documentation tells you to exhaust all alternatives before using Terraform provisioners to configure servers. The usual pattern is both: Terraform or OpenTofu to create the server, then Ansible, Salt or Puppet to configure it. Two licensing notes: Terraform 1.6 and later is under the Business Source License, while its fork OpenTofu is MPL-2.0. If you want the whole operating system described as code and rebuilt from it, NixOS takes a different approach entirely, but that means changing distributions.
Open source or commercial?
The licensing picture changed a lot in two years, so check what you'd actually run, not just the code's license. Ansible (GPL-3.0), Salt (Apache-2.0), pyinfra (MIT) and mgmt (GPL-3.0) are open source end to end. Puppet's code was Apache-2.0, but Perforce now develops Puppet Core as a proprietary product, with a free Developer license for fewer than 25 nodes; OpenVox is the open-source continuation. Chef's code is Apache-2.0, but Progress's binaries are under a EULA that's free only for personal, non-commercial use; Cinc is the free rebuild. CFEngine and Rudder are open core: a GPL-3.0 base with paid enterprise features. Commercial editions add support, web consoles, compliance content and, for Puppet, CVE fix SLAs. For a small Linux team, the open-source tools are complete; the paid editions earn their price when you need someone to call.
Do you need a configuration management tool?
If you run more than a couple of servers, yes, even a simple one. Without it, every server becomes a one-off: a package installed by hand here, a config line edited there, and nobody can rebuild it after a failure. A configuration management tool turns those steps into files you can review, version and run again on a new server in minutes. It also makes routine work safer, because a change goes through the same playbook or manifest every time, and check or noop modes show what would change first. Start small: put your SSH hardening, firewall rules, users and automatic updates into an Ansible playbook, run it in check mode, then for real. If you later need servers kept in line continuously, add an agent-based tool. For one hobby server, a short shell script in a git repository is a perfectly reasonable start.
How we checked this: On 14 Nov 2026 (UTC): Ansible (ansible-core 2.22.0) and pyinfra 3.11.1 ran from a workstation against a Debian 12.15 and an AlmaLinux 9.8 test VM, with nothing installed on the servers; Salt 3008.5 (salt-minion from the Salt Project's Broadcom-hosted repository) ran masterless with salt-call --local on the AlmaLinux VM; OpenVox 9.4.0 (openvox-agent from apt.voxpupuli.org) ran puppet apply --noop on the Debian VM. Everything was removed afterwards, including repositories, signing keys and users. Puppet, Chef, CFEngine, Rudder, Cinc and mgmt are researched from their documentation and were not run on our servers. Anything we didn't run ourselves comes from the official documentation listed under Sources.
Check that auto-patching is really working
unattended-upgrades-check runs 12 read-only checks over SSH on Ubuntu, Debian, AlmaLinux and Rocky Linux servers (unattended-upgrades or dnf-automatic) and gives each server a PASS, WARN or FAIL. Free and open source (Apache-2.0).
FAQ
What is the best alternative to SaltStack?
For most teams, Ansible: it does both configuration and ad-hoc remote commands, like Salt, but needs no agent or master, only SSH and Python on each server. If you liked Salt's agents keeping servers in line, OpenVox (free, Apache-2.0) or Puppet give you a pull-based agent that corrects drift on a schedule. pyinfra is the pick if you want to write Python instead of YAML and Jinja. Before switching, note that Salt itself is still maintained: three releases shipped on 26 October 2026.
What are the best Ansible alternatives?
pyinfra if you want the same agentless SSH model in plain Python. Salt if you need faster execution across large fleets or event-driven automation. Puppet or OpenVox, or Chef or Cinc, if you want agents that pull and enforce state continuously. CFEngine if you need a very light agent on many hosts.
Is SaltStack still maintained?
Yes. The Salt Project, now part of Broadcom, released 3008.5 LTS, 3006.30 LTS and 3007.17 on 26 October 2026. Its policy is one LTS release a year; 3008 LTS gets active support until June 2027 and security fixes until June 2028. Packages moved to packages.broadcom.com, and the code is still Apache-2.0.
Is Puppet still open source?
Its current builds aren't. Perforce moved new Puppet builds to a private location from early 2025, and its documentation says that since February 2026 it develops and supports Puppet Core 8 and 9 as a proprietary product. A Developer license is free for fewer than 25 nodes. The community fork OpenVox keeps publishing Apache-2.0 packages compatible with Puppet code.
Is Chef free to use?
The source code is Apache-2.0, but Progress's Chef binaries are distributed under the Chef EULA, and its free license is for personal and non-commercial use. Commercial SaaS plans start at $59 per node per year. Cinc rebuilds the same open-source code without the EULA, so you can use it in production for free, without support.
Is Terraform a configuration management tool?
No. Terraform, and its open-source fork OpenTofu, provision infrastructure through provider APIs: servers, networks, DNS, cloud services. Configuration management tools set up what runs inside a server. HashiCorp's own docs say to exhaust all alternatives before using Terraform's provisioners to configure servers. Most teams use both: Terraform or OpenTofu to create machines, then Ansible, Salt or Puppet to configure them.
Sources
- Ansible docs: installation guide (control and managed node requirements)
- ansible-core on PyPI (2.22.0)
- Red Hat Ansible Automation Platform pricing
- Salt Project blog (October 2026 releases)
- Salt install guide: version support lifecycle
- Salt install guide: supported operating systems
- Salt docs: Salt SSH
- Puppet pricing
- Puppet blog: open source Puppet updates for 2025
- Puppet docs: open source Puppet release notes (proprietary since February 2026)
- OpenVox by Vox Pupuli
- OpenVox install guide
- Chef pricing
- Chef docs: licensing
- Cinc: about the project
- CFEngine: Community vs Enterprise
- CFEngine docs: how CFEngine works
- Rudder pricing
- Rudder docs: supported operating systems
- pyinfra docs: compatibility
- mgmt on GitHub
- Terraform docs: provisioners