Your infrastructure. Your stack, not ours.

Muppy manages infrastructure the way you would by hand, but written down and replayable. It connects over SSH to the hosts you enrol, runs tasks on them and reports their state: a uniform control plane across the whole stack, from server to Kubernetes.

In production: Audaxis, Actiwork, Alter Eco Santé · Reference: La Poste

A uniform control plane

One foundation, applications you switch on as needed.

Tasks are published in addons installed when needed. Muppy Enterprise customers and partners can write their own.

Muppy Core

Hosts, tasks, facts

Physical servers, VMs or LXC containers enrolled over SSH, schedulable and traced Python tasks, systemd services, UFW firewall and SSH keys.

docs.muppy.io/muppy_core/
PostgreSQL

Backup, replication, HA, PITR

Four applications: clusters and backups, Replicated Cluster Set, automatic pglookout failover, to-the-minute restore with WAL-G.

docs.muppy.io/guides/postgresql/
Muppy K8s

Packages, releases, profiles

Muppy, Helm or Kubectl packages synced from git, reusable profiles, migrations with backup, multi-cloud Pack8s.

docs.muppy.io/muppy_k8s/
Monitoring

What should wake you up

Silent by default, never “OK” when it doesn’t know. Backups, replication, HA, PITR, Kubernetes certificates.

docs.muppy.io/monitoring/
Muppy Vault

Secrets encrypted at rest

Envelope encryption; the master key exists only in your password manager and in memory.

docs.muppy.io/vault/
In production

13 features that make the difference day to day.

🚨

Incidents & high availability

DF3 PostgreSQL locks

Activity and blocking locks detected and released, on a schedule if you like.

DF4 Smart Healthchecks

Programmable, topology-aware checks, with Pushover, Slack or Teams alerts.

DF6 400 GB+ filestore HA

Rsync every minute, 7 Prometheus metrics, an alert as soon as it stops.

DF7 Auto-refreshed CIDRs

Firewall rules kept in line with the IPs published by Cloudflare, GitLab, Azure DevOps.

DF8 Ephemeral firewall rules

Every UFW rule can carry an expiry date and disappears on its own.

🚀

Migrations & deployments

DF1 Database-aware upgrades

Stop, backup, migration, restart, traced database by database.

DF2 K8s Tag Mapper

A semantic git tag picks the Kubernetes profile and triggers the deployment.

DF5 Audited scheduled tasks

Every run linked to its log in the task queue, without SSH.

DF13 Two-wizard installation

Kubernetes, Traefik, cert-manager and PostgreSQL installed from the interface.

🔧

Updates & debugging

DF9 Browser IDE on any host

Code-Server on bare metal, a VM or an LXC container.

DF10 In-pod debugging

VS Code in the pod, with the same mounts and variables: the bug reproduces where it lives.

DF11 Dual URL + Sunray

A private route and a public route, protected by passkey with one switch.

DF12 K8s configuration log

The exact Helm command, console output and exit code of every install and upgrade.

Wherever your servers are

Cloud, multi-cloud, on-premise or hybrid.

🇪🇺

Sovereign providers

OVHScalewayInfomaniakIkoulaContaboDell ECS
🌍

Integrated hyperscalers

AWSAzureGoogle CloudCloudflareDigitalOcean
📡

Multi-provider DNS

OVHCloudflareScalewayMuppy built-in DNS

On OVH Public Cloud, Muppy creates the instances itself: they arrive enrolled and managed, with a decommission date as a safeguard. docs.muppy.io/ovh

PostgreSQL

How long you’re down, how much data you lose: you set both.

HA (with replication) answers the first, PITR the second. Together they cover a demanding RTO and an RPO of a few minutes, including with clusters spread across several clouds.

  • Automatic failover: a pglookout daemon on each cluster elects the best standby.
  • Restore to the minute: WAL shipped to S3 every minute, regular full backups.
  • And dumps: scheduled per-database backups, retention, scheduled restores.
RPO
min.

a few minutes of loss at most, with PITR

RTO
auto

the best standby promoted by pglookout

Reach
multi

clusters replicated across several clouds

Trace
100%

of HA events logged

Muppy Hoster

Your server estate, turned into a PaaS under your brand.

Muppy installs on your bare-metal servers or VPSs, with no migration. The Manganese layer isolates each customer in their own account, with their App Servers, and you stay in control of resources.

Existing estate keptNative multi-tenancyTurnkey CI/CDBuilt-in monitoringUFW, Sunray, Let’s Encrypt

Installing Muppy Enterprise

On your control server, bare metal or VPS.

Inventory of your estate

Enrolment of your hosts, SSH keys, DNS and certificates: Muppy models what you already have.

Activating Manganese

Each customer gets their account, their catalogue of App Definitions and their resource pool.

Your customers sign in

Through the Manganese interface or through MCP: they create their servers themselves.

Audaxis
4 months → 2 h

Time to set up a PostgreSQL infrastructure, as measured by Audaxis for a large enterprise customer.

Multi-DC PostgreSQL HAManaged KubernetesRCS replicationContinuous backup / PITROn-premise · SaaS · multi-cloud
They build Muppy into their offer

PostgreSQL superpowers for teams without DBAs.

An ERP integrator for more than 20 years (Contraste Europe group), Audaxis added Muppy to its catalogue to run highly available, multi-cloud PostgreSQL and Kubernetes infrastructures, with three services: audit, implementation, support and evolution.

Audaxis’s Muppy offer ↗ · Become a partner

Questions

What people ask us about Muppy

How is it different from Portainer or Ansible?

Portainer manages containers, Ansible automates configuration. Muppy covers the whole operations lifecycle in a single interface: hosts, PostgreSQL and its high availability, Kubernetes, DNS, TLS, CI/CD, monitoring and secrets, with a log of every task run.

Does Muppy work on existing infrastructure?

Yes. You enrol your existing servers (bare metal, VM or LXC); Muppy runs its tasks on them over SSH. It works with sovereign providers, hyperscalers and your own datacentres, and creates instances directly on OVH Public Cloud.

How does Muppy handle PostgreSQL high availability?

Through Replicated Cluster Set replication, automatic failover driven by pglookout (which can be switched off during maintenance), point-in-time recovery with WAL-G, and health checks that alert your phone.

Can Muppy be driven by an AI agent?

Yes. Muppy exposes an MCP server: the agent reads the same information you do, acts freely in development and test, and asks for human approval everywhere else.

Next step

Ready to take back control of your infrastructure?

In 30 minutes, we look at your servers, your databases and your availability goals, and propose a plan to take back control.

You describe your project

Four fields, no phone number required.

We talk it through for 30 minutes

Architecture, hosting constraints, security, taking over what you already run.

Prefer to try first? Try Manganese, free beta

Your project