deploy-ex-ops

name

deploy-ex-ops

description

Use when executing deploy_ex commands to operate on running infrastructure — deploying releases, uploading to S3, running Ansible deploys, managing QA nodes, autoscaling EC2 instances, running k6 load tests, SSH-ing into instances, checking instance health, viewing logs, connecting to remote IEx, rolling back releases, restarting apps, or checking what's currently deployed. This is for RUNNING existing deploy_ex mix tasks, not for writing new code or generating infrastructure files. Triggers on: deploy to production, upload release, rollback, QA node create/destroy, scale up/down, SSH into server, app is down, check health, restart app, view logs, iex console, load test, what's deployed, ping ansible hosts. Use this even without mentioning deploy_ex — if someone wants to deploy an Elixir app, check instance status, or restart a service in this project, this skill applies.

deploy_ex Operations

Guide for running deploy_ex deployment operations on Elixir projects targeting AWS.

Quick Reference

Release Cycle

mix deploy_ex.release [--only app1] [--force]   # build (change detection)
mix deploy_ex.upload [--parallel]                 # upload to S3
mix ansible.deploy [--only app1]                  # deploy to EC2

First-time Setup

mix deploy_ex.full_setup -ya     # -y auto-approve, -a auto-pull AWS creds (no deploy step; CI handles deploys)
mix deploy_ex.install_github_action

Rollback

mix ansible.rollback my_app [--select]   # --select for interactive history picker

Workflows

Building and Deploying a Release

deploy_ex detects which apps changed since the last release using git diff, mix.lock changes, and the mix deps.tree. Only changed apps get rebuilt.

1. mix deploy_ex.release — builds releases for changed apps. For Phoenix apps, runs the full asset pipeline (npm, esbuild, sass, tailwind, phx.digest) automatically.
2. mix deploy_ex.upload — uploads .tar.gz artifacts to S3. QA releases auto-detected from qa/* or qa-* branch names.
3. mix ansible.deploy — runs Ansible playbooks against EC2 instances. Use --target-sha abc1234 to deploy a specific SHA.

Force rebuild all: mix deploy_ex.release --force Target specific apps: --only app1 --only app2 or --except app3

QA Nodes

Ephemeral EC2 instances for testing specific SHAs:

mix deploy_ex.qa.create my_app --sha abc1234 [--attach-lb]
mix deploy_ex.qa.deploy my_app --sha def5678       # redeploy different SHA
mix deploy_ex.qa.attach_lb my_app                   # route traffic
mix deploy_ex.qa.detach_lb my_app                   # stop traffic
mix deploy_ex.qa.destroy my_app [--all]              # clean up
mix deploy_ex.qa.list                                # list nodes
mix deploy_ex.qa.cleanup                             # remove terminated from state

QA nodes reuse app-specific AMIs when available (skipping Ansible setup). State persisted to S3 at qa-nodes/{app_name}/{instance_id}.json.

Autoscaling

mix deploy_ex.autoscale.status my_app
mix deploy_ex.autoscale.scale my_app --desired 3
mix deploy_ex.autoscale.refresh my_app [-s Rolling] [-a launch-first] [-w]
mix deploy_ex.autoscale.refresh_status my_app

Strategies: Rolling (default, launch-then-terminate) or ReplaceRootVolume (in-place). Availability presets: launch-first (100%/110%, zero-downtime) or terminate-first (90%/100%, cost-saving).

Load Testing (k6)

mix deploy_ex.load_test.init my_app
mix deploy_ex.load_test.create_instance [--instance-type t3.small]
mix deploy_ex.load_test.upload --script load_test.js
mix deploy_ex.load_test.exec --target-url http://... [--prometheus-url http://...]
mix deploy_ex.load_test.destroy_instance

Connecting to Instances

mix deploy_ex.ssh my_app              # interactive SSH
mix deploy_ex.ssh my_app --root       # SSH as root
mix deploy_ex.ssh my_app --log        # stream app logs (journalctl)
mix deploy_ex.ssh my_app --iex        # remote IEx console
mix deploy_ex.ssh.authorize my_app    # add your SSH key

Instance Management

mix deploy_ex.instance.status my_app [-e prod]
mix deploy_ex.instance.health [--qa] [--all]
mix deploy_ex.restart_app my_app
mix deploy_ex.start_app my_app / mix deploy_ex.stop_app my_app
mix deploy_ex.restart_machine my_app
mix deploy_ex.find_nodes [--tag key=value] [--format table|json|ids]

Monitoring

mix deploy_ex.load_balancer.health my_app
mix deploy_ex.grafana.install_dashboard --id 19665    # k6 dashboard
mix deploy_ex.view_current_release my_app
mix deploy_ex.list_available_releases
mix deploy_ex.list_app_release_history my_app

Universal Options

Most tasks accept: --only (multi), --except (multi), --force/-f, --quiet/-q, --no-tui

Important Context

• All tasks require valid AWS credentials (env vars, AWS_PROFILE, or instance role)
• Tasks validate project type first via DeployExHelpers.check_valid_project() — works for both umbrella and single-app projects
• Destructive operations (drop, destroy) prompt for confirmation unless --force or --auto-approve
• TUI is auto-disabled in CI or when stdin isn't a TTY
• Release artifacts stored in S3 at {app}/{timestamp}-{sha}-{filename}.tar.gz

For full task reference, read docs/api-reference.md. For configuration, read docs/configuration-guide.md.