recipe-maintainer/.claude/commands/t1cc-start.md

description, allowed-tools

description	allowed-tools
Provision the t1cc DigitalOcean test server and deploy Traefik	Bash Read Write Edit Glob Grep

t1cc-start

Provision a fresh DigitalOcean droplet for t1cc (2 vCPU, 8 GB RAM), wait for cloud-init to finish, verify Docker Swarm health, and deploy Traefik. The reserved IP is static — DNS does not need updating.

Read and follow the instructions in .claude/commands/includes/logging.md.

Prerequisites

1. Check .testenv

Read terraform/.testenv. It must contain:

• DO_TOKEN — DigitalOcean API token
• RESERVED_IP — pre-allocated reserved IP (allocated on first run by setup.sh)

If DO_TOKEN is missing, stop and tell the user to create terraform/.testenv with their DO token:

echo 'DO_TOKEN=dop_v1_...' > terraform/.testenv

If RESERVED_IP is missing, that is fine — setup.sh will allocate one automatically and save it.

2. Check SSH key

Verify the SSH key exists:

test -f test-ssh/test-ssh-keys/nptest && echo "key exists" || echo "MISSING"

If the key is missing, stop and tell the user to place the private key at test-ssh/test-ssh-keys/nptest.

Steps

3. Run terraform

From the workspace root, run:

./terraform/setup.sh

This will:

• Allocate a reserved IP if not already present (and save it to .testenv)
• Write terraform/terraform.tfvars
• Run terraform init (if needed)
• Run terraform apply -auto-approve
• Wait for cloud-init to complete (handled by terraform's remote-exec provisioner)

This step can take 3–5 minutes. Stream the output and report progress. If it fails, show the terraform error and stop.

4. Verify server health via SSH

SSH into the server and run health checks. Use:

ssh -F test-ssh/ssh-config -o StrictHostKeyChecking=no t1cc.commoninternet.net "<command>"

Run these checks:

a. Docker is running:

ssh -F test-ssh/ssh-config -o StrictHostKeyChecking=no t1cc.commoninternet.net "docker info --format '{{.ServerVersion}}'"

b. Swarm is active:

ssh -F test-ssh/ssh-config -o StrictHostKeyChecking=no t1cc.commoninternet.net "docker info --format '{{.Swarm.LocalNodeState}}'"

Expected output: active

c. Proxy network exists:

ssh -F test-ssh/ssh-config -o StrictHostKeyChecking=no t1cc.commoninternet.net "docker network ls --filter name=proxy --format '{{.Name}}'"

Expected output: proxy

If any check fails, report the failure and stop. Do not proceed to Traefik deployment on an unhealthy server.

5. Deploy Traefik

Check whether the traefik app env file already exists:

test -f ~/.abra/servers/t1cc.commoninternet.net/t1cc.commoninternet.net.env && echo "exists" || echo "missing"

If the env file is missing — create the traefik app:

abra app new traefik --server t1cc.commoninternet.net --domain t1cc.commoninternet.net --no-input

Then set the required email:

abra app config t1cc.commoninternet.net --update LETS_ENCRYPT_EMAIL=certs@commoninternet.net

If the env file exists — check that LETS_ENCRYPT_EMAIL and DOMAIN are set correctly. Read the env file:

cat ~/.abra/servers/t1cc.commoninternet.net/t1cc.commoninternet.net.env | grep -E "LETS_ENCRYPT_EMAIL|^DOMAIN"

If DOMAIN is set to something other than t1cc.commoninternet.net, update it:

abra app config t1cc.commoninternet.net --update DOMAIN=t1cc.commoninternet.net

Deploy Traefik:

abra app deploy t1cc.commoninternet.net --chaos --force --no-input

Wait up to 90 seconds for the Traefik service to start, checking every 10 seconds:

ssh -F test-ssh/ssh-config -o StrictHostKeyChecking=no t1cc.commoninternet.net \
  "docker service ls --filter name=traefik --format '{{.Replicas}}'"

Wait until the replicas show 1/1. If not ready after 90 seconds, report a warning but continue.

6. Switch default instance to t1cc

python3 scripts/switch_default_instance.py t1cc

Verify:

python3 scripts/get_test_instance.py

7. Report

Print a summary:

• Reserved IP (from terraform/.testenv or terraform output -raw reserved_ip)
• Domain: t1cc.commoninternet.net
• Docker Swarm: active
• Traefik: running (1/1 replicas)
• Default instance: t1cc

Remind the user to run /init-instance if they want all maintained recipes deployed.