Disclaimer This is an ai made project only made to test the ai models capabilities, I have done a couple changes myself but nothing really major. The project does work but again you might prefer not to use it scince it is ai

sortie

Kubernetes for Rust.
Deploy, manage, and monitor Rust services across a cluster of servers from one CLI.

Features

Capability	What it does
Multi host clusters	Deploy to a fleet of servers, not just one
Rolling updates	Hosts updated one at a time with health gates
Canary & blue-green	Deploy to a subset, or spin a full parallel stack
Embedded reverse proxy	Built in HTTP proxy with Host based routing. No nginx needed.
Declarative state	Declare targets in sortie.toml; sortie apply reconciles
Auto rollback	Failed health checks trigger automatic rollback
Service registry	Register services, resolve by name, generate ingress config
Secrets management	Encrypted key-value store on servers (SSH is the boundary)
Auto scaling	CPU-based scaling loop, adjusts systemd instances
Self healing	sortie-agent watchdog restarts dead services
kubectl style CLI	apply, get, describe, logs, rollback, scale
Multi cluster federation	Deploy to all targets with one command
TLS provisioning	Auto certs via certbot (Let's Encrypt)
Observability	Per-host metrics (CPU, memory) and log aggregation
Environment config	Inline env vars for per-target service configuration

Quick start

cargo install sortie
sortie init

Edit sortie.toml with your server details, then deploy:

sortie apply production

This builds your binary, uploads it via SCP to every host, installs a systemd service, starts it, and waits for the health endpoint to respond. If a health check fails, that host rolls back.

To see what would change without deploying:

sortie apply production --check

Embedded proxy

No nginx, no caddy, no setup. Install the proxy on your server and it reads the service registry to route requests.

sortie svc register production my-api 8080
sortie proxy install production --port 80

The proxy matches the Host header against registered service names. If your server IP is 1.2.3.4:

curl -H "Host: my-api" http://1.2.3.4/

No domain needed. The service name you registered is the hostname clients send. You can use any string.

To add TLS, run sortie tls setup with a real domain first (Let's Encrypt), then point DNS at the server.

Canary and blue-green

Roll out to a subset of hosts before full deploy:

sortie apply production --canary 20

Or deploy a full parallel stack and flip:

sortie apply production --blue-green

Deploy to all targets

sortie deploy all

Iterates every target in sortie.toml. Also works with --canary and --blue-green.

Install

cargo install sortie

Or build from source:

git clone <repo-url>
cd sortie
cargo build --release

Requires Rust 1.70+, Linux servers with SSH + systemd, key-based auth.

Commands

Command	Like kubectl	Action
sortie init		Create a sortie.toml
sortie apply <target>	kubectl apply	Rolling deploy across all hosts. --check for dry-run, --canary/--blue-green for strategies
sortie deploy <target>		Same as apply, with --all for federation
sortie get	kubectl get	Show all targets with version, timestamp, services
sortie describe <target>	kubectl describe	Target config + per-host status
sortie logs <target> [host]	kubectl logs	Fetch service logs from a host
sortie rollback <target>	kubectl rollout undo	Revert to previous binary
sortie status <target>	kubectl get pods	Check service status per host
sortie health <target>		Check host connectivity and service health
sortie svc register <target> <name> <port>		Register a service in the cluster
sortie svc list <target>		List all registered services
sortie svc resolve <target> <name>		Resolve a service name to host:port
sortie svc restart <host> <target> <name>	kubectl rollout restart	Restart a service on a host
sortie svc stop <host> <target> <name>		Stop a service on a host
sortie svc start <host> <target> <name>		Start a service on a host
sortie proxy install <target>		Install embedded reverse proxy as a systemd service
sortie scale <target> <n>	kubectl scale	Set systemd instance count per host
sortie autoscale <target>		Start CPU-based auto-scaling loop
sortie secret set <target> <key> <value>		Store a secret file on the server
sortie secret get <target> <key>		Read a stored secret
sortie secret rm <target> <key>		Delete a stored secret
sortie tls setup <target> <domain> <email>		Provision Let's Encrypt certs via certbot
sortie ingress <target>		Generate nginx config from service registry
sortie metrics <target>		Show CPU and memory per host

Configuration

[targets.production]
hosts = ["10.0.0.1", "10.0.0.2"]
port = 22
user = "deploy"
key_path = "~/.ssh/id_rsa"
target_triple = "x86_64-unknown-linux-gnu"
deploy_path = "/opt/myapp"
health_check_url = "http://localhost:8080/health"
health_check_timeout_secs = 30
pre_deploy = "cd /opt/myapp && ./run_migrations.sh"
post_deploy = "echo 'Deployment completed'"

[targets.production.service]
name = "myapp"
restart = "always"

[targets.production.env]
DATABASE_URL = "postgres://localhost:5432/myapp"
RUST_LOG = "info"

Configuration reference

Field	Required	Default	Description
hosts	yes		Server addresses in the cluster
port	no	22	SSH port
user	yes		SSH user
key_path	no	~/.ssh/id_rsa	SSH key path
target_triple	yes		Rust target triple
deploy_path	yes		Server directory for binary
health_check_url	no		URL to check after deploy
health_check_timeout_secs	no	30	Max wait for health
instances	no	1	Expected replicas per host
pre_deploy	no		Command to run on each host before the new binary takes traffic
post_deploy	no		Command to run on each host after the deploy succeeds
service.name	yes		systemd service name
service.restart	no	always	Restart policy
env.*	no		Environment variables

How it works

Reads desired state from sortie.toml. Builds the binary. Performs a rolling deployment across all hosts: upload via SCP, install systemd unit with env vars, verify health, next host. If a health check fails, that host rolls back and the deploy stops.

Previous versions are kept as .bak files for instant rollback. Cluster state is stored in .sortie/state.json on each host.

Architecture

Project structure

sortie/
  Cargo.toml
  Cargo.lock
  LICENSE
  README.md
  CHANGELOG.md
  CONTRIBUTING.md
  CODE_OF_CONDUCT.md
  SECURITY.md
  SECURITY_CONTACTS
  SUPPORT.md
  OWNERS
  OWNERS_ALIASES
  .gitattributes
  .gitignore
  rust-toolchain.toml
  Makefile
  cmd/
    sortie/
      main.rs
    agent/
      main.rs
  pkg/
    lib.rs
    cli.rs
    types.rs
    config.rs
    init.rs
    build.rs
    ssh.rs
    cluster.rs
    deploy.rs
    systemd.rs
    health.rs
    rollback.rs
    proxy.rs
  logo/
    sortie.png
  build/
    build-release.bash
  hack/
    ci.bash
  tests/
    smoke_test.rs
  docs/
    architecture/
      images/
        c4-context--0.svg
        c4-dynamic-build--0.svg
        build-lifecycle--0.svg
        c4-containers--0.svg
        c4-components--0.svg
        c4-deployment--0.svg
  LICENSES/
    README.md
  examples/
    basic.toml
    multi-host.toml
    with-env.toml
  third_party/
    README.md
  CHANGELOG/
    v0.1.0.md
    v0.2.0.md

Contributing

Pull requests welcome. See CONTRIBUTING.md.

License

MIT. See LICENSE.