Skip to main content

Cargo SSH Jobs

Cargo SSH is an advanced Switchboard workflow for inspectable Script runtime jobs.

Use the normal Node.js webserver path first. Cargo SSH is useful when you want an Acurast Shell job that exposes Dropbear through Switchboard's public SNI route and uses the Acurast Cargo bridge signer for privileged registration and certificate requests.

Scaffold​

mkdir -p switchboard-ssh-demo
proof switchboard init \
  --template ssh \
  --distro ubuntu \
  --project-dir ./switchboard-ssh-demo \
  --context mainnet

The template generates an inspectable Script/Cargo project. Its bootstrap uses Dropbear for SSH and the Cargo bridge signer for:

  • deployment-intent claim
  • job-signed ingress registration
  • job-ACME certificate request signing

It does not fall back to JOB_SIGNER_PRIVATE_KEY inside the job.

Deploy​

cd switchboard-ssh-demo
proof switchboard preflight --quote
proof switchboard deploy --yes --dry-run --json
proof switchboard deploy --yes

Script runtime deploys require Acurast processors that advertise the Shell module. Switchboard filters capacity for that runtime, but live availability can still change between preflight, quote, and submit.

Connect Through SNI​

After deployment, use the hostname from the deploy report. The public route is TLS/SNI passthrough to the job, which then exposes the SSH banner.

If you are diagnosing the route, start with:

proof switchboard deploy doctor --report <report.json> --probe

--probe is read-only. It performs public TLS/SNI and SSH banner checks. It does not spend, deploy, mutate DNS, mutate routes, or record settlement.

Bridge Diagnostics​

Generated jobs include:

switchboard-cargo-bridge-doctor

Run it from an SSH login shell when PROOF support asks you to prove the job's Cargo bridge shape. It signs a fixed diagnostic challenge and avoids printing the bridge socket, private key, token, seed, bearer token, or full reusable signature material.

Common Issues​

  • No Shell processor available: try later or ask PROOF for supported capacity.
  • Public probe resets: check whether the job advertised the processor LAN address and whether gateway route-state points at the job TLS port.
  • Login shell lacks BRIDGE_SOCKET: this can be normal. Original job processes may still retain bridge access; use the bridge doctor helper.
  • Late local timeout: run deploy status and deploy doctor before spending again.