Bump version for 0.13.2

Revert "Fix typo in fixture filename"

.github/workflows/ci.yml

@@ -1,5 +1,9 @@
 name: CI
-on: [push, pull_request]
+on: 
+  push:
+    branches:
+      - main
+  pull_request:
 jobs:
   tests:
     strategy:

CONTRIBUTING.md

@@ -27,7 +27,7 @@ Please keep the following guidelines in mind when opening a pull request:
 - Add tests for your changes, if possible.
 - Ensure that your changes don't break existing functionality.
 
-#### Commit message guidline
+#### Commit message guidelines
 
 A good commit message should describe what changed and why.
 

Dockerfile

@@ -14,14 +14,14 @@ COPY Gemfile Gemfile.lock mrsk.gemspec ./
 COPY lib/mrsk/version.rb /mrsk/lib/mrsk/version.rb
 
 # Install system dependencies
-RUN apk add --no-cache --update build-base git docker openrc \
+RUN apk add --no-cache --update build-base git docker openrc openssh-client-default \
     && rc-update add docker boot \
     && gem install bundler --version=2.4.3 \
     && bundle install
 
 # Copy the rest of our application code into the container.
 # We do this after bundle install, to avoid having to run bundle
-# everytime we do small fixes in the source code.
+# every time we do small fixes in the source code.
 COPY . .
 
 # Install the gem locally from the project folder
@@ -31,6 +31,10 @@ RUN gem build mrsk.gemspec && \
 # Set the working directory to /workdir
 WORKDIR /workdir
 
+# Tell git it's safe to access /workdir/.git even if
+# the directory is owned by a different user
+RUN git config --global --add safe.directory /workdir
+
 # Set the entrypoint to run the installed binary in /workdir
 # Example:  docker run -it -v "$PWD:/workdir" mrsk init
 ENTRYPOINT ["mrsk"]

Gemfile.lock

@@ -1,11 +1,12 @@
 PATH
   remote: .
   specs:
-    mrsk (0.10.0)
+    mrsk (0.13.2)
       activesupport (>= 7.0)
       bcrypt_pbkdf (~> 1.0)
       dotenv (~> 2.8)
       ed25519 (~> 1.2)
+      net-ssh (~> 7.0)
       sshkit (~> 1.21)
       thor (~> 1.2)
       zeitwerk (~> 2.5)
@@ -35,13 +36,18 @@ GEM
     builder (3.2.4)
     concurrent-ruby (1.2.2)
     crass (1.0.6)
-    debug (1.7.1)
+    debug (1.7.2)
+      irb (>= 1.5.0)
+      reline (>= 0.3.1)
     dotenv (2.8.1)
     ed25519 (1.3.0)
     erubi (1.12.0)
     i18n (1.12.0)
       concurrent-ruby (~> 1.0)
-    loofah (2.19.1)
+    io-console (0.6.0)
+    irb (1.6.3)
+      reline (>= 0.3.0)
+    loofah (2.20.0)
       crass (~> 1.0.2)
       nokogiri (>= 1.5.9)
     method_source (1.0.0)
@@ -74,6 +80,8 @@ GEM
       thor (~> 1.0)
       zeitwerk (~> 2.5)
     rake (13.0.6)
+    reline (0.3.3)
+      io-console (~> 0.5)
     ruby2_keywords (0.0.5)
     sshkit (1.21.4)
       net-scp (>= 1.1.2)

README.md

@@ -6,6 +6,8 @@ Watch the screencast: https://www.youtube.com/watch?v=LL1cV2FXZ5I
 
 Join us on Discord: https://discord.gg/YgHVT7GCXS
 
+Ask questions: https://github.com/mrsked/mrsk/discussions
+
 ## Installation
 
 If you have a Ruby environment available, you can install MRSK globally with:
@@ -14,13 +16,13 @@ If you have a Ruby environment available, you can install MRSK globally with:
 gem install mrsk
 ```
 
-...otherwise, you can run a dockerized version via an alias (add this to your ${SHELL}rc to simplify re-use):
+...otherwise, you can run a dockerized version via an alias (add this to your .bashrc or similar to simplify re-use):
 
 ```sh
 alias mrsk='docker run --rm -it -v $HOME/.ssh:/root/.ssh -v /var/run/docker.sock:/var/run/docker.sock -v ${PWD}/:/workdir  ghcr.io/mrsked/mrsk'
 ```
 
-Then, inside your app directory, run `mrsk init` (or `mrsk init --bundle` within Rails apps where you want a bin/mrsk binstub). Now edit the new file `config/deploy.yml`. It could look as simple as this:
+Then, inside your app directory, run `mrsk init` (or `mrsk init --bundle` within Rails 7+ apps where you want a bin/mrsk binstub). Now edit the new file `config/deploy.yml`. It could look as simple as this:
 
 ```yaml
 service: hey
@@ -42,30 +44,30 @@ Then edit your `.env` file to add your registry password as `MRSK_REGISTRY_PASSW
 Now you're ready to deploy to the servers:
 
 ```
-mrsk deploy
+mrsk setup
 ```
 
 This will:
 
 1. Connect to the servers over SSH (using root by default, authenticated by your ssh key)
-2. Install Docker on any server that might be missing it (using apt-get): root access is needed via ssh for this.
+2. Install Docker and curl on any server that might be missing it (using apt-get): root access is needed via ssh for this.
 3. Log into the registry both locally and remotely
 4. Build the image using the standard Dockerfile in the root of the application.
 5. Push the image to the registry.
 6. Pull the image from the registry onto the servers.
 7. Ensure Traefik is running and accepting traffic on port 80.
-8. Ensure your app responds with `200 OK` to `GET /up`.
+8. Ensure your app responds with `200 OK` to `GET /up` (you must have curl installed inside your app image!).
 9. Start a new container with the version of the app that matches the current git version hash.
 10. Stop the old container running the previous version of the app.
 11. Prune unused images and stopped containers to ensure servers don't fill up.
 
-Voila! All the servers are now serving the app on port 80. If you're just running a single server, you're ready to go. If you're running multiple servers, you need to put a load balancer in front of them.
+Voila! All the servers are now serving the app on port 80. If you're just running a single server, you're ready to go. If you're running multiple servers, you need to put a load balancer in front of them. For subsequent deploys, or if your servers already have Docker and curl installed, you can just run `mrsk deploy`.
 
 ## Vision
 
 In the past decade+, there's been an explosion in commercial offerings that make deploying web apps easier. Heroku kicked it off with an incredible offering that stayed ahead of the competition seemingly forever. These days we have excellent alternatives like Fly.io and Render. And hosted Kubernetes is making things easier too on AWS, GCP, Digital Ocean, and elsewhere. But these are all offerings that have you renting computers in the cloud at a premium. If you want to run on your own hardware, or even just have a clear migration path to do so in the future, you need to carefully consider how locked in you get to these commercial platforms. Preferably before the bills swallow your business whole!
 
-MRSK seeks to bring the advance in ergonomics pioneered by these commercial offerings to deploying web apps anywhere. Whether that's low-cost cloud options without the managed-service markup from the likes of Digital Ocean, Hetzner, OVH, etc, or it's your own colocated bare metal. To MRSK, it's all the same. Feed the config file a list of IP addresses with vanilla Ubuntu servers that have seen no prep beyond an added SSH key, and you'll be running in literally minutes.
+MRSK seeks to bring the advance in ergonomics pioneered by these commercial offerings to deploying web apps anywhere. Whether that's low-cost cloud options without the managed-service markup from the likes of Digital Ocean, Hetzner, OVH, etc., or it's your own colocated bare metal. To MRSK, it's all the same. Feed the config file a list of IP addresses with vanilla Ubuntu servers that have seen no prep beyond an added SSH key, and you'll be running in literally minutes.
 
 This approach gives you enormous portability. You can have your web app deployed on several clouds at ease like this. Or you can buy the baseline with your own hardware, then deploy to a cloud before a big seasonal spike to get more capacity. When you're not locked into a single provider from a tooling perspective, there are a lot of compelling options available.
 
@@ -182,6 +184,19 @@ registry:
 
 A reference to secret `DOCKER_REGISTRY_TOKEN` will look for `ENV["DOCKER_REGISTRY_TOKEN"]` on the machine running MRSK.
 
+#### Using AWS ECR as the container registry
+
+AWS ECR's access token is only valid for 12hrs. In order to not have to manually regenerate the token every time, you can use ERB in the `deploy.yml` file to shell out to the `aws` cli command, and obtain the token:
+
+```yaml
+registry:
+  server: <your aws account id>.dkr.ecr.<your aws region id>.amazonaws.com
+  username: AWS
+  password: <%= %x(aws ecr get-login-password) %>
+```
+
+You will need to have the `aws` CLI installed locally for this to work.
+
 ### Using a different SSH user than root
 
 The default SSH user is root, but you can change it using `ssh/user`:
@@ -191,6 +206,15 @@ ssh:
   user: app
 ```
 
+If you are using non-root user, you need to bootstrap your servers manually, before using them with MRSK. On Ubuntu, you'd do:
+
+```bash
+sudo apt update
+sudo apt upgrade -y
+sudo apt install -y docker.io curl git
+sudo usermod -a -G docker ubuntu
+```
+
 ### Using a proxy SSH host
 
 If you need to connect to server through a proxy host, you can use `ssh/proxy`:
@@ -207,6 +231,13 @@ ssh:
   proxy: "app@192.168.0.1"
 ```
 
+Also if you need specific proxy command to connect to the server:
+
+```yaml
+ssh:
+  proxy_command: aws ssm start-session --target %h --document-name AWS-StartSSHSession --parameters 'portNumber=%p' --region=us-east-1 ## ssh via aws ssm
+```
+
 ### Using env variables
 
 You can inject env variables into the app containers using `env`:
@@ -288,8 +319,9 @@ You can specialize the default Traefik rules by setting labels on the containers
 
 ```yaml
 labels:
-  traefik.http.routers.hey.rule: Host(`app.hey.com`)
+  traefik.http.routers.hey-web.rule: Host(`app.hey.com`)
 ```
+Traefik rules are in the "service-role-destination" format. The default role will be `web` if no rule is specified. If the destination is not specified, it is not included. To give an example, the above rule would become "traefik.http.routers.hey-web-staging.rule" if it was for the "staging" destination.
 
 Note: The backticks are needed to ensure the rule is passed in correctly and not treated as command substitution by Bash!
 
@@ -312,6 +344,21 @@ servers:
       my-label: "50"
 ```
 
+### Using shell expansion
+
+You can use shell expansion to interpolate values from the host machine into labels and env variables with the `${}` syntax.
+Anything within the curly braces will be executed on the host machine and the result will be interpolated into the label or env variable.
+
+```yaml
+labels:
+  host-machine: "${cat /etc/hostname}"
+
+env:
+  HOST_DEPLOYMENT_DIR: "${PWD}"
+```
+
+Note: Any other occurrence of `$` will be escaped to prevent unwanted shell expansion!
+
 ### Using container options
 
 You can specialize the options used to start containers using the `options` definitions:
@@ -439,9 +486,9 @@ RUN --mount=type=secret,id=GITHUB_TOKEN \
   rm -rf /usr/local/bundle/cache
 ```
 
-### Using command arguments for Traefik
+### Traefik command arguments
 
-You can customize the traefik command line:
+Customize the Traefik command line using `args`:
 
 ```yaml
 traefik:
@@ -450,37 +497,70 @@ traefik:
     accesslog.format: json
 ```
 
-This will start the traefik container with `--accesslog=true accesslog.format=json`.
+This starts the Traefik container with `--accesslog=true --accesslog.format=json` arguments.
 
-### Traefik's host port binding
+### Traefik host port binding
 
-By default Traefik binds to port 80 of the host machine, it can be configured to use an alternative port:
+Traefik binds to port 80 by default. Specify an alternative port using `host_port`:
 
 ```yaml
 traefik:
   host_port: 8080
 ```
 
-### Configure docker options for traefik
+### Traefik version, upgrades, and custom images
 
-We allow users to pass additional docker options to the trafik container like
+MRSK runs the traefik:v2.9 image to track Traefik 2.9.x releases.
+
+To pin Traefik to a specific version or an image published to your registry,
+specify `image`:
+
+```yaml
+traefik:
+  image: traefik:v2.10.0-rc1
+```
+
+This is useful for downgrading Traefik if there's an unexpected breaking
+change in a minor version release, upgrading Traefik to test forthcoming
+releases, or running your own Traefik-derived image.
+
+MRSK has not been tested for compatibility with Traefik 3 betas. Please do!
+
+### Traefik container configuration
+
+Pass additional Docker configuration for the Traefik container using `options`:
 
 ```yaml
 traefik:
   options:
     publish:
       - 8080:8080
-    volumes:
+    volume:
       - /tmp/example.json:/tmp/example.json
     memory: 512m
 ```
 
-This will start the traefik container with a command like: `docker run ... --volume /tmp/example.json:/tmp/example.json --publish 8080:8080 `
+This starts the Traefik container with `--volume /tmp/example.json:/tmp/example.json --publish 8080:8080 --memory 512m` arguments to `docker run`.
 
+### Traefik container labels
 
-### Configure alternate entrypoints for traefik
+Add labels to Traefik Docker container.
 
-You can configure multiple entrypoints for traefik like so:
+```yaml
+traefik:
+  labels:
+    traefik.enable: true
+    traefik.http.routers.dashboard.rule: Host(`traefik.example.com`) && (PathPrefix(`/api`) || PathPrefix(`/dashboard`))
+    traefik.http.routers.dashboard.service: api@internal
+    traefik.http.routers.dashboard.middlewares: auth
+    traefik.http.middlewares.auth.basicauth.users: test:$2y$05$H2o72tMaO.TwY1wNQUV1K.fhjRgLHRDWohFvUZOJHBEtUXNKrqUKi # test:password
+```
+
+This labels Traefik container with `--label traefik.http.routers.dashboard.middlewares=\"auth\"` and so on.
+
+### Traefik alternate entrypoints
+
+You can configure multiple entrypoints for Traefik like so:
 
 ```yaml
 service: myservice
@@ -540,7 +620,7 @@ accessories:
       memory: "2GB"
   redis:
     image: redis:latest
-    role:
+    roles:
       - web
     port: "36379:6379"
     volumes:
@@ -588,40 +668,46 @@ servers:
 
 This assumes the Cron settings are stored in `config/crontab`.
 
-### Using audit broadcasts
+### Healthcheck
 
-If you'd like to broadcast audits of deploys, rollbacks, etc to a chatroom or elsewhere, you can configure the `audit_broadcast_cmd` setting with the path to a bin file that will be passed the audit line as the first argument:
+MRSK uses Docker healthchecks to check the health of your application during deployment. Traefik uses this same healthcheck status to determine when a container is ready to receive traffic.
 
-```yaml
-audit_broadcast_cmd:
-  bin/audit_broadcast
-```
-
-The broadcast command could look something like:
-
-```bash
-#!/usr/bin/env bash
-curl -q -d content="[My App] ${1}" https://3.basecamp.com/XXXXX/integrations/XXXXX/buckets/XXXXX/chats/XXXXX/lines
-```
-
-That'll post a line like follows to a preconfigured chatbot in Basecamp:
-
-```
-[My App] [dhh] Rolled back to version d264c4e92470ad1bd18590f04466787262f605de
-```
-
-### Using custom healthcheck path or port
-
-MRSK defaults to checking the health of your application again `/up` on port 3000. You can tailor both with the `healthcheck` setting:
+The healthcheck defaults to testing the HTTP response to the path `/up` on port 3000, up to 7 times. You can tailor this behaviour with the `healthcheck` setting:
 
 ```yaml
 healthcheck:
   path: /healthz
   port: 4000
+  max_attempts: 7
+  interval: 20s
 ```
 
 This will ensure your application is configured with a traefik label for the healthcheck against `/healthz` and that the pre-deploy healthcheck that MRSK performs is done against the same path on port 4000.
 
+You can also specify a custom healthcheck command, which is useful for non-HTTP services:
+
+```yaml
+healthcheck:
+  cmd: /bin/check_health
+```
+
+The top-level healthcheck configuration applies to all services that use
+Traefik, by default. You can also specialize the configuration at the role
+level:
+
+```yaml
+servers:
+  job:
+    hosts: ...
+    cmd: bin/jobs
+    healthcheck:
+      cmd: bin/check
+```
+
+The healthcheck allows for an optional `max_attempts` setting, which will attempt the healthcheck up to the specified number of times before failing the deploy. This is useful for applications that take a while to start up. The default is 7.
+
+Note: The HTTP health checks assume that the `curl` command is available inside the container. If that's not the case, use the healthcheck's `cmd` option to specify an alternative check that the container supports.
+
 ## Commands
 
 ### Running commands on servers
@@ -754,13 +840,86 @@ Message: Automatic deploy lock
 You can also manually acquire and release the lock
 
 ```
-mrsk lock acquire -m "Doing maintanence"
+mrsk lock acquire -m "Doing maintenance"
 ```
 
 ```
 mrsk lock release
 ```
 
+## Rolling deployments
+
+When deploying to large numbers of hosts, you might prefer not to restart your services on every host at the same time.
+
+MRSK's default is to boot new containers on all hosts in parallel. But you can control this by configuring `boot/limit` and `boot/wait` as options:
+
+```yaml
+service: myservice
+
+boot:
+  limit: 10 # Can also specify as a percentage of total hosts, such as "25%"
+  wait: 2
+```
+
+When `limit` is specified, containers will be booted on, at most, `limit` hosts at once. MRSK will pause for `wait` seconds between batches.
+
+These settings only apply when booting containers (using `mrsk deploy`, or `mrsk app boot`). For other commands, MRSK continues to run commands in parallel across all hosts.
+
+## Hooks
+
+You can run custom scripts at specific points with hooks.
+
+Hooks should be stored in the .mrsk/hooks folder. Running mrsk init will build that folder and add some sample scripts.
+
+You can change their location by setting `hooks_path` in the configuration file.
+
+If the script returns a non-zero exit code the command will be aborted.
+
+`MRSK_*` environment variables are available to the hooks command for
+fine-grained audit reporting, e.g. for triggering deployment reports or
+firing a JSON webhook. These variables include:
+- `MRSK_RECORDED_AT` - UTC timestamp in ISO 8601 format, e.g. `2023-04-14T17:07:31Z`
+- `MRSK_PERFORMER` - the local user performing the command (from `whoami`)
+- `MRSK_SERVICE_VERSION` - an abbreviated service and version for use in messages, e.g. app@150b24f
+- `MRSK_VERSION` - an full version being deployed
+- `MRSK_HOSTS` - a comma separated list of the hosts targeted by the command
+- `MRSK_COMMAND` - The command we are running
+- `MRSK_SUBCOMMAND` - optional: The subcommand we are running
+- `MRSK_DESTINATION` - optional: destination, e.g. "staging"
+- `MRSK_ROLE` - optional: role targeted, e.g. "web"
+
+There are four hooks:
+
+1. pre-connect
+Called before taking the deploy lock. For checks that need to run before connecting to remote hosts - e.g. DNS warming.
+
+2. pre-build
+Used for pre-build checks - e.g. there are no uncommitted changes or that CI has passed.
+
+3. pre-deploy
+For final checks before deploying, e.g. checking CI completed
+
+3. post-deploy - run after a deploy, redeploy or rollback
+
+This hook is also passed a `MRSK_RUNTIME` env variable.
+
+This could be used to broadcast a deployment message, or register the new version with an APM.
+
+The command could look something like:
+
+```bash
+#!/usr/bin/env bash
+curl -q -d content="[My App] ${MRSK_PERFORMER} Rolled back to version ${MRSK_VERSION}" https://3.basecamp.com/XXXXX/integrations/XXXXX/buckets/XXXXX/chats/XXXXX/lines
+```
+
+That'll post a line like follows to a preconfigured chatbot in Basecamp:
+
+```
+[My App] [dhh] Rolled back to version d264c4e92470ad1bd18590f04466787262f605de
+```
+
+Set `--skip_hooks` to avoid running the hooks.
+
 ## Stage of development
 
 This is beta software. Commands may still move around. But we're live in production at [37signals](https://37signals.com).

bin/mrsk

@@ -8,9 +8,11 @@ require "mrsk"
 begin
   Mrsk::Cli::Main.start(ARGV)
 rescue SSHKit::Runner::ExecuteError => e
-  puts "  \e[31mERROR (#{e.cause.class}): #{e.cause.message}\e[0m"
+  puts "  \e[31mERROR (#{e.cause.class}): #{e.message}\e[0m"
   puts e.cause.backtrace if ENV["VERBOSE"]
+  exit 1
 rescue => e
   puts "  \e[31mERROR (#{e.class}): #{e.message}\e[0m"
   puts e.backtrace if ENV["VERBOSE"]
+  exit 1
 end

lib/mrsk/cli.rb

@@ -1,4 +1,6 @@
 module Mrsk::Cli
+  class LockError < StandardError; end
+  class HookError < StandardError; end
 end
 
 # SSHKit uses instance eval, so we need a global const for ergonomics

lib/mrsk/cli/accessory.rb

@@ -14,8 +14,6 @@ class Mrsk::Cli::Accessory < Mrsk::Cli::Base
             execute *MRSK.auditor.record("Booted #{name} accessory"), verbosity: :debug
             execute *accessory.run
           end
-
-          audit_broadcast "Booted accessory #{name}" unless options[:skip_broadcast]
         end
       end
     end

lib/mrsk/cli/app.rb

@@ -2,30 +2,38 @@ class Mrsk::Cli::App < Mrsk::Cli::Base
   desc "boot", "Boot app on servers (or reboot app if already running)"
   def boot
     with_lock do
+      hold_lock_on_error do
         say "Get most recent version available as an image...", :magenta unless options[:version]
         using_version(version_or_latest) do |version|
           say "Start container with version #{version} using a #{MRSK.config.readiness_delay}s readiness delay (or reboot if already running)...", :magenta
 
-        cli = self
+          on(MRSK.hosts) do
+            execute *MRSK.auditor.record("Tagging #{MRSK.config.absolute_image} as the latest image"), verbosity: :debug
+            execute *MRSK.app.tag_current_as_latest
+          end
 
-        on(MRSK.hosts) do |host|
+          on(MRSK.hosts, **MRSK.boot_strategy) do |host|
             roles = MRSK.roles_on(host)
 
             roles.each do |role|
-            execute *MRSK.auditor(role: role).record("Booted app version #{version}"), verbosity: :debug
+              app = MRSK.app(role: role)
+              auditor = MRSK.auditor(role: role)
 
-            begin
-              if capture_with_info(*MRSK.app(role: role).container_id_for_version(version)).present?
-                tmp_version = "#{version}_#{SecureRandom.hex(8)}"
+              if capture_with_info(*app.container_id_for_version(version, only_running: true), raise_on_non_zero_exit: false).present?
+                tmp_version = "#{version}_replaced_#{SecureRandom.hex(8)}"
                 info "Renaming container #{version} to #{tmp_version} as already deployed on #{host}"
-                execute *MRSK.auditor(role: role).record("Renaming container #{version} to #{tmp_version}"), verbosity: :debug
-                execute *MRSK.app(role: role).rename_container(version: version, new_version: tmp_version)
+                execute *auditor.record("Renaming container #{version} to #{tmp_version}"), verbosity: :debug
+                execute *app.rename_container(version: version, new_version: tmp_version)
               end
 
-              old_version = capture_with_info(*MRSK.app(role: role).current_running_version).strip
-              execute *MRSK.app(role: role).run
-              sleep MRSK.config.readiness_delay
-              execute *MRSK.app(role: role).stop(version: old_version), raise_on_non_zero_exit: false if old_version.present?
+              execute *auditor.record("Booted app version #{version}"), verbosity: :debug
+
+              old_version = capture_with_info(*app.current_running_version, raise_on_non_zero_exit: false).strip
+              execute *app.start_or_run(hostname: "#{host}-#{SecureRandom.hex(6)}")
+
+              Mrsk::Utils::HealthcheckPoller.wait_for_healthy(pause_after_ready: true) { capture_with_info(*app.status(version: version)) }
+
+              execute *app.stop(version: old_version), raise_on_non_zero_exit: false if old_version.present?
             end
           end
         end
@@ -54,7 +62,7 @@ class Mrsk::Cli::App < Mrsk::Cli::Base
         roles = MRSK.roles_on(host)
 
         roles.each do |role|
-          execute *MRSK.auditor(role: role).record("Stopped app"), verbosity: :debug
+          execute *MRSK.auditor.record("Stopped app", role: role), verbosity: :debug
           execute *MRSK.app(role: role).stop, raise_on_non_zero_exit: false
         end
       end
@@ -101,7 +109,7 @@ class Mrsk::Cli::App < Mrsk::Cli::Base
           roles = MRSK.roles_on(host)
 
           roles.each do |role|
-            execute *MRSK.auditor(role: role).record("Executed cmd '#{cmd}' on app version #{version}"), verbosity: :debug
+            execute *MRSK.auditor.record("Executed cmd '#{cmd}' on app version #{version}", role: role), verbosity: :debug
             puts_by_host host, capture_with_info(*MRSK.app(role: role).execute_in_existing_container(cmd))
           end
         end
@@ -124,6 +132,31 @@ class Mrsk::Cli::App < Mrsk::Cli::Base
     on(MRSK.hosts) { |host| puts_by_host host, capture_with_info(*MRSK.app.list_containers) }
   end
 
+  desc "stale_containers", "Detect app stale containers"
+  option :stop, aliases: "-s", type: :boolean, default: false, desc: "Stop the stale containers found"
+  def stale_containers
+    with_lock do
+      stop = options[:stop]
+
+      cli = self
+
+      on(MRSK.hosts) do |host|
+        roles = MRSK.roles_on(host)
+
+        roles.each do |role|
+          cli.send(:stale_versions, host: host, role: role).each do |version|
+            if stop
+              puts_by_host host, "Stopping stale container for role #{role} with version #{version}"
+              execute *MRSK.app(role: role).stop(version: version), raise_on_non_zero_exit: false
+            else
+              puts_by_host host,  "Detected stale container for role #{role} with version #{version} (use `mrsk app stale_containers --stop` to stop)"
+            end
+          end
+        end
+      end
+    end
+  end
+
   desc "images", "Show app images on servers"
   def images
     on(MRSK.hosts) { |host| puts_by_host host, capture_with_info(*MRSK.app.list_images) }
@@ -183,7 +216,7 @@ class Mrsk::Cli::App < Mrsk::Cli::Base
         roles = MRSK.roles_on(host)
 
         roles.each do |role|
-          execute *MRSK.auditor(role: role).record("Removed app container with version #{version}"), verbosity: :debug
+          execute *MRSK.auditor.record("Removed app container with version #{version}", role: role), verbosity: :debug
           execute *MRSK.app(role: role).remove_container(version: version)
         end
       end
@@ -197,7 +230,7 @@ class Mrsk::Cli::App < Mrsk::Cli::Base
         roles = MRSK.roles_on(host)
 
         roles.each do |role|
-          execute *MRSK.auditor(role: role).record("Removed all app containers"), verbosity: :debug
+          execute *MRSK.auditor.record("Removed all app containers", role: role), verbosity: :debug
           execute *MRSK.app(role: role).remove_containers
         end
       end
@@ -240,6 +273,17 @@ class Mrsk::Cli::App < Mrsk::Cli::Base
       version.presence
     end
 
+    def stale_versions(host:, role:)
+      versions = nil
+      on(host) do
+        versions = \
+          capture_with_info(*MRSK.app(role: role).list_versions, raise_on_non_zero_exit: false)
+          .split("\n")
+          .drop(1)
+      end
+      versions
+    end
+
     def version_or_latest
       options[:version] || "latest"
     end

lib/mrsk/cli/base.rb

@@ -6,8 +6,6 @@ module Mrsk::Cli
   class Base < Thor
     include SSHKit::DSL
 
-    class LockError < StandardError; end
-
     def self.exit_on_failure?() true end
 
     class_option :verbose, type: :boolean, aliases: "-v", desc: "Detailed logging"
@@ -22,7 +20,7 @@ module Mrsk::Cli
     class_option :config_file, aliases: "-c", default: "config/deploy.yml", desc: "Path to config file"
     class_option :destination, aliases: "-d", desc: "Specify destination to be used for config file (staging -> deploy.staging.yml)"
 
-    class_option :skip_broadcast, aliases: "-B", type: :boolean, default: false, desc: "Skip audit broadcasts"
+    class_option :skip_hooks, aliases: "-H", type: :boolean, default: false, desc: "Don't run hooks"
 
     def initialize(*)
       super
@@ -74,38 +72,100 @@ module Mrsk::Cli
         puts "  Finished all in #{sprintf("%.1f seconds", runtime)}"
       end
 
-      def audit_broadcast(line)
-        run_locally { execute *MRSK.auditor.broadcast(line), verbosity: :debug }
-      end
-
       def with_lock
+        if MRSK.holding_lock?
+          yield
+        else
+          run_hook "pre-connect"
+
           acquire_lock
 
+          begin
             yield
-      ensure
+          rescue
+            if MRSK.hold_lock_on_error?
+              error "  \e[31mDeploy lock was not released\e[0m"
+            else
               release_lock
             end
 
-      def acquire_lock
-        if MRSK.lock_count == 0
-          say "Acquiring the deploy lock"
-          on(MRSK.primary_host) { execute *MRSK.lock.acquire("Automatic deploy lock", MRSK.config.version) }
-        end
-        MRSK.lock_count += 1
-      rescue SSHKit::Runner::ExecuteError => e
-        if e.message =~ /cannot create directory/
-          invoke "mrsk:cli:lock:status", []
+            raise
           end
 
-        raise LockError, "Deploy lock found"
+          release_lock
+        end
+      end
+
+      def acquire_lock
+        raise_if_locked do
+          say "Acquiring the deploy lock...", :magenta
+          on(MRSK.primary_host) { execute *MRSK.lock.acquire("Automatic deploy lock", MRSK.config.version), verbosity: :debug }
+        end
+
+        MRSK.holding_lock = true
       end
 
       def release_lock
-        MRSK.lock_count -= 1
-        if MRSK.lock_count == 0
-          say "Releasing the deploy lock"
-          on(MRSK.primary_host) { execute *MRSK.lock.release }
+        say "Releasing the deploy lock...", :magenta
+        on(MRSK.primary_host) { execute *MRSK.lock.release, verbosity: :debug }
+
+        MRSK.holding_lock = false
       end
+
+      def raise_if_locked
+        yield
+      rescue SSHKit::Runner::ExecuteError => e
+        if e.message =~ /cannot create directory/
+          on(MRSK.primary_host) { puts capture_with_debug(*MRSK.lock.status) }
+          raise LockError, "Deploy lock found"
+        else
+          raise e
+        end
+      end
+
+      def hold_lock_on_error
+        if MRSK.hold_lock_on_error?
+          yield
+        else
+          MRSK.hold_lock_on_error = true
+          yield
+          MRSK.hold_lock_on_error = false
+        end
+      end
+
+      def run_hook(hook, **extra_details)
+        if !options[:skip_hooks] && MRSK.hook.hook_exists?(hook)
+          details = { hosts: MRSK.hosts.join(","), command: command, subcommand: subcommand }
+
+          say "Running the #{hook} hook...", :magenta
+          run_locally do
+            MRSK.with_verbosity(:debug) { execute *MRSK.hook.run(hook, **details, **extra_details) }
+          rescue SSHKit::Command::Failed
+            raise HookError.new("Hook `#{hook}` failed")
+          end
+        end
+      end
+
+      def command
+        @mrsk_command ||= begin
+          invocation_class, invocation_commands = *first_invocation
+          if invocation_class == Mrsk::Cli::Main
+            invocation_commands[0]
+          else
+            Mrsk::Cli::Main.subcommand_classes.find { |command, clazz| clazz == invocation_class }[0]
+          end
+        end
+      end
+
+      def subcommand
+        @mrsk_subcommand ||= begin
+          invocation_class, invocation_commands = *first_invocation
+          invocation_commands[0] if invocation_class != Mrsk::Cli::Main
+        end
+      end
+
+      def first_invocation
+        instance_variable_get("@_invocations").first
       end
     end
 end

lib/mrsk/cli/build.rb

@@ -1,4 +1,6 @@
 class Mrsk::Cli::Build < Mrsk::Cli::Base
+  class BuildError < StandardError; end
+
   desc "deliver", "Build app and push app image to registry then pull image on servers"
   def deliver
     with_lock do
@@ -12,6 +14,9 @@ class Mrsk::Cli::Build < Mrsk::Cli::Base
     with_lock do
       cli = self
 
+      verify_local_dependencies
+      run_hook "pre-build"
+
       run_locally do
         begin
           MRSK.with_verbosity(:debug) { execute *MRSK.builder.push }
@@ -77,4 +82,19 @@ class Mrsk::Cli::Build < Mrsk::Cli::Base
       puts capture(*MRSK.builder.info)
     end
   end
+
+  private
+    def verify_local_dependencies
+      run_locally do
+        begin
+          execute *MRSK.builder.ensure_local_dependencies_installed
+        rescue SSHKit::Command::Failed => e
+          build_error = e.message =~ /command not found/ ?
+            "Docker is not installed locally" :
+            "Docker buildx plugin is not installed locally"
+
+          raise BuildError, build_error
+        end
+      end
+    end
 end

lib/mrsk/cli/healthcheck.rb

@@ -1,8 +1,4 @@
 class Mrsk::Cli::Healthcheck < Mrsk::Cli::Base
-  MAX_ATTEMPTS = 7
-
-  class HealthcheckError < StandardError; end
-
   default_command :perform
 
   desc "perform", "Health check current app version"
@@ -10,37 +6,11 @@ class Mrsk::Cli::Healthcheck < Mrsk::Cli::Base
     on(MRSK.primary_host) do
       begin
         execute *MRSK.healthcheck.run
-
-        target = "Health check against #{MRSK.config.healthcheck["path"]}"
-        attempt = 1
-
-        begin
-          status = capture_with_info(*MRSK.healthcheck.curl)
-
-          if status == "200"
-            info "#{target} succeeded with 200 OK!"
-          else
-            raise HealthcheckError, "#{target} failed with status #{status}"
-          end
-        rescue SSHKit::Command::Failed
-          if attempt <= MAX_ATTEMPTS
-            info "#{target} failed to respond, retrying in #{attempt}s..."
-            sleep attempt
-            attempt += 1
-
-            retry
-          else
-            raise
-          end
-        end
-      rescue SSHKit::Command::Failed, HealthcheckError => e
+        Mrsk::Utils::HealthcheckPoller.wait_for_healthy { capture_with_info(*MRSK.healthcheck.status) }
+      rescue Mrsk::Utils::HealthcheckPoller::HealthcheckError => e
         error capture_with_info(*MRSK.healthcheck.logs)
-
-        if e.message =~ /curl/
-          raise SSHKit::Command::Failed, "#{target} failed to return 200 OK!"
-        else
+        error capture_with_pretty_json(*MRSK.healthcheck.container_health_log)
         raise
-        end
       ensure
         execute *MRSK.healthcheck.stop, raise_on_non_zero_exit: false
         execute *MRSK.healthcheck.remove, raise_on_non_zero_exit: false

lib/mrsk/cli/lock.rb

@@ -2,25 +2,25 @@ class Mrsk::Cli::Lock < Mrsk::Cli::Base
   desc "status", "Report lock status"
   def status
     handle_missing_lock do
-      on(MRSK.primary_host) { puts capture_with_info(*MRSK.lock.status) }
+      on(MRSK.primary_host) { puts capture_with_debug(*MRSK.lock.status) }
     end
   end
 
   desc "acquire", "Acquire the deploy lock"
-  option :message, aliases: "-m", type: :string, desc: "A lock mesasge", required: true
+  option :message, aliases: "-m", type: :string, desc: "A lock message", required: true
   def acquire
     message = options[:message]
-    handle_missing_lock do
-      on(MRSK.primary_host) { execute *MRSK.lock.acquire(message, MRSK.config.version) }
-      say "Set the deploy lock"
+    raise_if_locked do
+      on(MRSK.primary_host) { execute *MRSK.lock.acquire(message, MRSK.config.version), verbosity: :debug }
+      say "Acquired the deploy lock"
     end
   end
 
   desc "release", "Release the deploy lock"
   def release
     handle_missing_lock do
-      on(MRSK.primary_host) { execute *MRSK.lock.release }
-      say "Removed the deploy lock"
+      on(MRSK.primary_host) { execute *MRSK.lock.release, verbosity: :debug }
+      say "Released the deploy lock"
     end
   end
 

lib/mrsk/cli/main.rb

@@ -1,8 +1,8 @@
 class Mrsk::Cli::Main < Mrsk::Cli::Base
   desc "setup", "Setup all accessories and deploy app to servers"
   def setup
-    with_lock do
     print_runtime do
+      with_lock do
         invoke "mrsk:cli:server:bootstrap"
         invoke "mrsk:cli:accessory:boot", [ "all" ]
         deploy
@@ -13,13 +13,10 @@ class Mrsk::Cli::Main < Mrsk::Cli::Base
   desc "deploy", "Deploy app to servers"
   option :skip_push, aliases: "-P", type: :boolean, default: false, desc: "Skip image build and push"
   def deploy
+    runtime = print_runtime do
       with_lock do
         invoke_options = deploy_options
 
-      runtime = print_runtime do
-        say "Ensure curl and Docker are installed...", :magenta
-        invoke "mrsk:cli:server:bootstrap", [], invoke_options
-
         say "Log into image registry...", :magenta
         invoke "mrsk:cli:registry:login", [], invoke_options
 
@@ -31,29 +28,34 @@ class Mrsk::Cli::Main < Mrsk::Cli::Base
           invoke "mrsk:cli:build:deliver", [], invoke_options
         end
 
+        run_hook "pre-deploy"
+
         say "Ensure Traefik is running...", :magenta
         invoke "mrsk:cli:traefik:boot", [], invoke_options
 
         say "Ensure app can pass healthcheck...", :magenta
         invoke "mrsk:cli:healthcheck:perform", [], invoke_options
 
+        say "Detect stale containers...", :magenta
+        invoke "mrsk:cli:app:stale_containers", [], invoke_options
+
         invoke "mrsk:cli:app:boot", [], invoke_options
 
         say "Prune old containers and images...", :magenta
         invoke "mrsk:cli:prune:all", [], invoke_options
       end
-
-      audit_broadcast "Deployed #{service_version} in #{runtime.round} seconds" unless options[:skip_broadcast]
     end
+
+    run_hook "post-deploy", runtime: runtime.round
   end
 
   desc "redeploy", "Deploy app to servers without bootstrapping servers, starting Traefik, pruning, and registry login"
   option :skip_push, aliases: "-P", type: :boolean, default: false, desc: "Skip image build and push"
   def redeploy
+    runtime = print_runtime do
       with_lock do
         invoke_options = deploy_options
 
-      runtime = print_runtime do
         if options[:skip_push]
           say "Pull app image...", :magenta
           invoke "mrsk:cli:build:pull", [], invoke_options
@@ -62,46 +64,45 @@ class Mrsk::Cli::Main < Mrsk::Cli::Base
           invoke "mrsk:cli:build:deliver", [], invoke_options
         end
 
+        run_hook "pre-deploy"
+
         say "Ensure app can pass healthcheck...", :magenta
         invoke "mrsk:cli:healthcheck:perform", [], invoke_options
 
+        say "Detect stale containers...", :magenta
+        invoke "mrsk:cli:app:stale_containers", [], invoke_options
+
         invoke "mrsk:cli:app:boot", [], invoke_options
       end
-
-      audit_broadcast "Redeployed #{service_version} in #{runtime.round} seconds" unless options[:skip_broadcast]
     end
+
+    run_hook "post-deploy", runtime: runtime.round
   end
 
   desc "rollback [VERSION]", "Rollback app to VERSION"
   def rollback(version)
+    rolled_back = false
+    runtime = print_runtime do
       with_lock do
+        invoke_options = deploy_options
+
         MRSK.config.version = version
-
-      if container_name_available?(MRSK.config.service_with_version)
-        say "Start version #{version}, then wait #{MRSK.config.readiness_delay}s for app to boot before stopping the old version...", :magenta
-
-        cli = self
         old_version = nil
 
-        on(MRSK.hosts) do |host|
-          old_version = capture_with_info(*MRSK.app.current_running_version).strip.presence
+        if container_available?(version)
+          run_hook "pre-deploy"
 
-          execute *MRSK.app.start
-
-          if old_version
-            sleep MRSK.config.readiness_delay
-
-            execute *MRSK.app.stop(version: old_version), raise_on_non_zero_exit: false
-          end
-        end
-
-        audit_broadcast "Rolled back #{service_version(Mrsk::Utils.abbreviate_version(old_version))} to #{service_version}" unless options[:skip_broadcast]
+          invoke "mrsk:cli:app:boot", [], invoke_options.merge(version: version)
+          rolled_back = true
         else
           say "The app version '#{version}' is not available as a container (use 'mrsk app containers' for available versions)", :red
         end
       end
     end
 
+    run_hook "post-deploy", runtime: runtime.round if rolled_back
+  end
+
   desc "details", "Show details about all containers"
   def details
     invoke "mrsk:cli:traefik:details"
@@ -119,7 +120,7 @@ class Mrsk::Cli::Main < Mrsk::Cli::Base
   desc "config", "Show combined config (including secrets!)"
   def config
     run_locally do
-      puts MRSK.config.to_h.to_yaml
+      puts Mrsk::Utils.redacted(MRSK.config.to_h).to_yaml
     end
   end
 
@@ -141,6 +142,14 @@ class Mrsk::Cli::Main < Mrsk::Cli::Base
       puts "Created .env file"
     end
 
+    unless (hooks_dir = Pathname.new(File.expand_path(".mrsk/hooks"))).exist?
+      hooks_dir.mkpath
+      Pathname.new(File.expand_path("templates/sample_hooks", __dir__)).each_child do |sample_hook|
+        FileUtils.cp sample_hook, hooks_dir, preserve: true
+      end
+      puts "Created sample hooks in .mrsk/hooks"
+    end
+
     if options[:bundle]
       if (binstub = Pathname.new(File.expand_path("bin/mrsk"))).exist?
         puts "Binstub already exists in bin/mrsk (remove first to create a new one)"
@@ -198,6 +207,9 @@ class Mrsk::Cli::Main < Mrsk::Cli::Base
   desc "healthcheck", "Healthcheck application"
   subcommand "healthcheck", Mrsk::Cli::Healthcheck
 
+  desc "lock", "Manage the deploy lock"
+  subcommand "lock", Mrsk::Cli::Lock
+
   desc "prune", "Prune old application images and containers"
   subcommand "prune", Mrsk::Cli::Prune
 
@@ -210,21 +222,28 @@ class Mrsk::Cli::Main < Mrsk::Cli::Base
   desc "traefik", "Manage Traefik load balancer"
   subcommand "traefik", Mrsk::Cli::Traefik
 
-  desc "lock", "Manage the deploy lock"
-  subcommand "lock", Mrsk::Cli::Lock
-
   private
-    def container_name_available?(container_name, host: MRSK.primary_host)
-      container_names = nil
-      on(host) { container_names = capture_with_info(*MRSK.app.list_container_names).split("\n") }
-      Array(container_names).include?(container_name)
+    def container_available?(version)
+      begin
+        on(MRSK.hosts) do
+          MRSK.roles_on(host).each do |role|
+            container_id = capture_with_info(*MRSK.app(role: role).container_id_for_version(version))
+            raise "Container not found" unless container_id.present?
+          end
+        end
+      rescue SSHKit::Runner::ExecuteError => e
+        if e.message =~ /Container not found/
+          say "Error looking for container version #{version}: #{e.message}"
+          return false
+        else
+          raise
+        end
+      end
+
+      true
     end
 
     def deploy_options
       { "version" => MRSK.config.version }.merge(options.without("skip_push"))
     end
-
-    def service_version(version = MRSK.config.abbreviated_version)
-      [ MRSK.config.service, version ].compact.join("@")
-    end
 end

lib/mrsk/cli/prune.rb

@@ -7,17 +7,18 @@ class Mrsk::Cli::Prune < Mrsk::Cli::Base
     end
   end
 
-  desc "images", "Prune unused images older than 7 days"
+  desc "images", "Prune dangling images"
   def images
     with_lock do
       on(MRSK.hosts) do
         execute *MRSK.auditor.record("Pruned images"), verbosity: :debug
-        execute *MRSK.prune.images
+        execute *MRSK.prune.dangling_images
+        execute *MRSK.prune.tagged_images
       end
     end
   end
 
-  desc "containers", "Prune stopped containers older than 3 days"
+  desc "containers", "Prune all stopped containers, except the last 5"
   def containers
     with_lock do
       on(MRSK.hosts) do

lib/mrsk/cli/server.rb

@@ -1,17 +1,21 @@
 class Mrsk::Cli::Server < Mrsk::Cli::Base
-  desc "bootstrap", "Ensure curl and Docker are installed on servers"
+  desc "bootstrap", "Set up Docker to run MRSK apps"
   def bootstrap
-    with_lock do
-      on(MRSK.hosts + MRSK.accessory_hosts) do
-        dependencies_to_install = Array.new.tap do |dependencies|
-          dependencies << "curl" unless execute "which curl", raise_on_non_zero_exit: false
-          dependencies << "docker.io" unless execute "which docker", raise_on_non_zero_exit: false
+    missing = []
+
+    on(MRSK.hosts | MRSK.accessory_hosts) do |host|
+      unless execute(*MRSK.docker.installed?, raise_on_non_zero_exit: false)
+        if execute(*MRSK.docker.superuser?, raise_on_non_zero_exit: false)
+          info "Missing Docker on #{host}. Installing…"
+          execute *MRSK.docker.install
+        else
+          missing << host
+        end
+      end
     end
 
-        if dependencies_to_install.any?
-          execute "apt-get update -y && apt-get install #{dependencies_to_install.join(" ")} -y"
-        end
-      end
+    if missing.any?
+      raise "Docker is not installed on #{missing.join(", ")} and can't be automatically installed without having root access and the `curl` command available. Install Docker manually: https://docs.docker.com/engine/install/"
     end
   end
 end

lib/mrsk/cli/templates/deploy.yml

@@ -25,10 +25,6 @@ registry:
 #   secret:
 #     - RAILS_MASTER_KEY
 
-# Call a broadcast command on deploys.
-# audit_broadcast_cmd:
-#   bin/broadcast_to_bc
-
 # Use a different ssh user than root
 # ssh:
 #   user: app

lib/mrsk/cli/templates/sample_hooks/post-deploy.sample

@@ -0,0 +1,14 @@
+#!/bin/sh
+
+# A sample post-deploy hook
+#
+# These environment variables are available:
+# MRSK_RECORDED_AT
+# MRSK_PERFORMER
+# MRSK_VERSION
+# MRSK_HOSTS
+# MRSK_ROLE (if set)
+# MRSK_DESTINATION (if set)
+# MRSK_RUNTIME
+
+echo "$MRSK_PERFORMER deployed $MRSK_VERSION to $MRSK_DESTINATION in $MRSK_RUNTIME seconds"

lib/mrsk/cli/templates/sample_hooks/pre-build.sample

@@ -0,0 +1,51 @@
+#!/bin/sh
+
+# A sample pre-build hook
+#
+# Checks:
+# 1. We have a clean checkout
+# 2. A remote is configured
+# 3. The branch has been pushed to the remote
+# 4. The version we are deploying matches the remote
+#
+# These environment variables are available:
+# MRSK_RECORDED_AT
+# MRSK_PERFORMER
+# MRSK_VERSION
+# MRSK_HOSTS
+# MRSK_ROLE (if set)
+# MRSK_DESTINATION (if set)
+
+if [ -n "$(git status --porcelain)" ]; then
+  echo "Git checkout is not clean, aborting..." >&2
+  git status --porcelain >&2
+  exit 1
+fi
+
+first_remote=$(git remote)
+
+if [ -z "$first_remote" ]; then
+  echo "No git remote set, aborting..." >&2
+  exit 1
+fi
+
+current_branch=$(git branch --show-current)
+
+if [ -z "$current_branch" ]; then
+  echo "No git remote set, aborting..." >&2
+  exit 1
+fi
+
+remote_head=$(git ls-remote $first_remote --tags $current_branch | cut -f1)
+
+if [ -z "$remote_head" ]; then
+  echo "Branch not pushed to remote, aborting..." >&2
+  exit 1
+fi
+
+if [ "$MRSK_VERSION" != "$remote_head" ]; then
+  echo "Version ($MRSK_VERSION) does not match remote HEAD ($remote_head), aborting..." >&2
+  exit 1
+fi
+
+exit 0

lib/mrsk/cli/templates/sample_hooks/pre-connect.sample

@@ -0,0 +1,47 @@
+#!/usr/bin/env ruby
+
+# A sample pre-connect check
+#
+# Warms DNS before connecting to hosts in parallel
+#
+# These environment variables are available:
+# MRSK_RECORDED_AT
+# MRSK_PERFORMER
+# MRSK_VERSION
+# MRSK_HOSTS
+# MRSK_ROLE (if set)
+# MRSK_DESTINATION (if set)
+# MRSK_RUNTIME
+
+hosts = ENV["MRSK_HOSTS"].split(",")
+results = nil
+max = 3
+
+elapsed = Benchmark.realtime do
+  results = hosts.map do |host|
+    Thread.new do
+      tries = 1
+
+      begin
+        Socket.getaddrinfo(host, 0, Socket::AF_UNSPEC, Socket::SOCK_STREAM, nil, Socket::AI_CANONNAME)
+      rescue SocketError
+        if tries < max
+          puts "Retrying DNS warmup: #{host}"
+          tries += 1
+          sleep rand
+          retry
+        else
+          puts "DNS warmup failed: #{host}"
+          host
+        end
+      end
+
+      tries
+    end
+  end.map(&:value)
+end
+
+retries = results.sum - hosts.size
+nopes = results.count { |r| r == max }
+
+puts "Prewarmed %d DNS lookups in %.2f sec: %d retries, %d failures" % [ hosts.size, elapsed, retries, nopes ]

lib/mrsk/cli/templates/sample_hooks/pre-deploy.sample

@@ -0,0 +1,82 @@
+#!/bin/sh
+
+# A sample pre-deploy hook
+#
+# Checks the Github status of the build, waiting for a pending build to complete for up to 720 seconds.
+#
+# Fails unless the combined status is "success"
+#
+# These environment variables are available:
+# MRSK_RECORDED_AT
+# MRSK_PERFORMER
+# MRSK_VERSION
+# MRSK_HOSTS
+# MRSK_COMMAND
+# MRSK_SUBCOMMAND
+# MRSK_ROLE (if set)
+# MRSK_DESTINATION (if set)
+
+#!/usr/bin/env ruby
+
+# Only check the build status for production deployments
+if ENV["MRSK_COMMAND"] == "rollback" || ENV["MRSK_DESTINATION"] != "production"
+  exit 0
+end
+
+require "bundler/inline"
+
+# true = install gems so this is fast on repeat invocations
+gemfile(true, quiet: true) do
+  source "https://rubygems.org"
+
+  gem "octokit"
+  gem "faraday-retry"
+end
+
+MAX_ATTEMPTS = 72
+ATTEMPTS_GAP = 10
+
+def exit_with_error(message)
+  $stderr.puts message
+  exit 1
+end
+
+def first_status_url(combined_status, state)
+  first_status = combined_status[:statuses].find { |status| status[:state] == state }
+  first_status && first_status[:target_url]
+end
+
+remote_url = `git config --get remote.origin.url`.strip.delete_prefix("https://github.com/")
+git_sha = `git rev-parse HEAD`.strip
+
+repository = Octokit::Repository.from_url(remote_url)
+github_client = Octokit::Client.new(access_token: ENV["GITHUB_TOKEN"])
+attempts = 0
+
+begin
+  loop do
+    combined_status = github_client.combined_status(remote_url, git_sha)
+    state = combined_status[:state]
+    first_status_url = first_status_url(combined_status, state)
+
+    case state
+    when "success"
+      puts "Build passed, see #{first_status_url}"
+      exit 0
+    when "failure"
+      exit_with_error "Build failed, see #{first_status_url}"
+    when "pending"
+      attempts += 1
+    end
+
+    puts "Waiting #{ATTEMPTS_GAP} more seconds for build to complete#{", see #{first_status_url}" if first_status_url}..."
+
+    if attempts == MAX_ATTEMPTS
+      exit_with_error "Build status is still pending, gave up after #{MAX_ATTEMPTS * ATTEMPTS_GAP} seconds"
+    end
+
+    sleep(ATTEMPTS_GAP)
+  end
+rescue Octokit::NotFound
+  exit_with_error "Build status could not be found"
+end

lib/mrsk/cli/traefik.rb

@@ -2,7 +2,10 @@ class Mrsk::Cli::Traefik < Mrsk::Cli::Base
   desc "boot", "Boot Traefik on servers"
   def boot
     with_lock do
-      on(MRSK.traefik_hosts) { execute *MRSK.traefik.run, raise_on_non_zero_exit: false }
+      on(MRSK.traefik_hosts) do
+        execute *MRSK.registry.login
+        execute *MRSK.traefik.run, raise_on_non_zero_exit: false
+      end
     end
   end
 
@@ -91,7 +94,7 @@ class Mrsk::Cli::Traefik < Mrsk::Cli::Base
     end
   end
 
-  desc "remove_container", "Remove Traefik image from servers", hide: true
+  desc "remove_image", "Remove Traefik image from servers", hide: true
   def remove_image
     with_lock do
       on(MRSK.traefik_hosts) do

lib/mrsk/commander.rb

@@ -2,11 +2,12 @@ require "active_support/core_ext/enumerable"
 require "active_support/core_ext/module/delegation"
 
 class Mrsk::Commander
-  attr_accessor :verbosity, :lock_count
+  attr_accessor :verbosity, :holding_lock, :hold_lock_on_error
 
   def initialize
     self.verbosity = :info
-    self.lock_count = 0
+    self.holding_lock = false
+    self.hold_lock_on_error = false
   end
 
   def config
@@ -35,7 +36,7 @@ class Mrsk::Commander
   end
 
   def primary_host
-    specific_hosts&.first || config.primary_web_host
+    specific_hosts&.first || specific_roles&.first&.primary_host || config.primary_web_host
   end
 
   def roles
@@ -50,6 +51,14 @@ class Mrsk::Commander
     end
   end
 
+  def boot_strategy
+    if config.boot.limit.present?
+      { in: :groups, limit: config.boot.limit, wait: config.boot.wait }
+    else
+      {}
+    end
+  end
+
   def roles_on(host)
     roles.select { |role| role.hosts.include?(host.to_s) }.map(&:name)
   end
@@ -75,18 +84,30 @@ class Mrsk::Commander
     Mrsk::Commands::Accessory.new(config, name: name)
   end
 
-  def auditor(role: nil)
-    Mrsk::Commands::Auditor.new(config, role: role)
+  def auditor(**details)
+    Mrsk::Commands::Auditor.new(config, **details)
   end
 
   def builder
     @builder ||= Mrsk::Commands::Builder.new(config)
   end
 
+  def docker
+    @docker ||= Mrsk::Commands::Docker.new(config)
+  end
+
   def healthcheck
     @healthcheck ||= Mrsk::Commands::Healthcheck.new(config)
   end
 
+  def hook
+    @hook ||= Mrsk::Commands::Hook.new(config)
+  end
+
+  def lock
+    @lock ||= Mrsk::Commands::Lock.new(config)
+  end
+
   def prune
     @prune ||= Mrsk::Commands::Prune.new(config)
   end
@@ -99,10 +120,6 @@ class Mrsk::Commander
     @traefik ||= Mrsk::Commands::Traefik.new(config)
   end
 
-  def lock
-    @lock ||= Mrsk::Commands::Lock.new(config)
-  end
-
   def with_verbosity(level)
     old_level = self.verbosity
 
@@ -115,6 +132,14 @@ class Mrsk::Commander
     SSHKit.config.output_verbosity = old_level
   end
 
+  def holding_lock?
+    self.holding_lock
+  end
+
+  def hold_lock_on_error?
+    self.hold_lock_on_error
+  end
+
   private
     # Lazy setup of SSHKit
     def configure_sshkit_with(config)

lib/mrsk/commands/app.rb

@@ -1,4 +1,6 @@
 class Mrsk::Commands::App < Mrsk::Commands::Base
+  ACTIVE_DOCKER_STATUSES = [ :running, :restarting ]
+
   attr_reader :role
 
   def initialize(config, role: nil)
@@ -6,15 +8,21 @@ class Mrsk::Commands::App < Mrsk::Commands::Base
     @role = role
   end
 
-  def run
+  def start_or_run(hostname: nil)
+    combine start, run(hostname: hostname), by: "||"
+  end
+
+  def run(hostname: nil)
     role = config.role(self.role)
 
     docker :run,
       "--detach",
       "--restart unless-stopped",
       "--name", container_name,
+      *(["--hostname", hostname] if hostname),
       "-e", "MRSK_CONTAINER_NAME=\"#{container_name}\"",
       *role.env_args,
+      *role.health_check_args,
       *config.logging_args,
       *config.volume_args,
       *role.label_args,
@@ -27,9 +35,13 @@ class Mrsk::Commands::App < Mrsk::Commands::Base
     docker :start, container_name
   end
 
+  def status(version:)
+    pipe container_id_for_version(version), xargs(docker(:inspect, "--format", DOCKER_HEALTH_STATUS_FORMAT))
+  end
+
   def stop(version: nil)
     pipe \
-      version ? container_id_for_version(version) : current_container_id,
+      version ? container_id_for_version(version) : current_running_container_id,
       xargs(config.stop_wait_time ? docker(:stop, "-t", config.stop_wait_time) : docker(:stop))
   end
 
@@ -40,7 +52,7 @@ class Mrsk::Commands::App < Mrsk::Commands::Base
 
   def logs(since: nil, lines: nil, grep: nil)
     pipe \
-      current_container_id,
+      current_running_container_id,
       "xargs docker logs#{" --since #{since}" if since}#{" --tail #{lines}" if lines} 2>&1",
       ("grep '#{grep}'" if grep)
   end
@@ -48,7 +60,7 @@ class Mrsk::Commands::App < Mrsk::Commands::Base
   def follow_logs(host:, grep: nil)
     run_over_ssh \
       pipe(
-        current_container_id,
+        current_running_container_id,
         "xargs docker logs --timestamps --tail 10 --follow 2>&1",
         (%(grep "#{grep}") if grep)
       ),
@@ -82,20 +94,23 @@ class Mrsk::Commands::App < Mrsk::Commands::Base
   end
 
 
-  def current_container_id
-    docker :ps, "--quiet", *filter_args
+  def current_running_container_id
+    docker :ps, "--quiet", *filter_args(statuses: ACTIVE_DOCKER_STATUSES), "--latest"
   end
 
-  def container_id_for_version(version)
-    container_id_for(container_name: container_name(version))
+  def container_id_for_version(version, only_running: false)
+    container_id_for(container_name: container_name(version), only_running: only_running)
   end
 
   def current_running_version
-    # FIXME: Find more graceful way to extract the version from "app-version" than using sed and tail!
+    list_versions("--latest", statuses: ACTIVE_DOCKER_STATUSES)
+  end
+
+  def list_versions(*docker_args, statuses: nil)
     pipe \
-      docker(:ps, *filter_args, "--format", '"{{.Names}}"'),
-      %(sed 's/-/\\n/g'),
-      "tail -n 1"
+      docker(:ps, *filter_args(statuses: statuses), *docker_args, "--format", '"{{.Names}}"'),
+      %(grep -oE "\\-[^-]+$"), # Extract SHA from "service-role-dest-SHA"
+      %(cut -c 2-)
   end
 
   def list_containers
@@ -128,20 +143,27 @@ class Mrsk::Commands::App < Mrsk::Commands::Base
     docker :image, :prune, "--all", "--force", *filter_args
   end
 
+  def tag_current_as_latest
+    docker :tag, config.absolute_image, config.latest_image
+  end
+
 
   private
     def container_name(version = nil)
       [ config.service, role, config.destination, version || config.version ].compact.join("-")
     end
 
-    def filter_args
-      argumentize "--filter", filters
+    def filter_args(statuses: nil)
+      argumentize "--filter", filters(statuses: statuses)
     end
 
-    def filters
+    def filters(statuses: nil)
       [ "label=service=#{config.service}" ].tap do |filters|
         filters << "label=destination=#{config.destination}" if config.destination
         filters << "label=role=#{role}" if role
+        statuses&.each do |status|
+          filters << "status=#{status}"
+        end
       end
     end
 end

lib/mrsk/commands/auditor.rb

@@ -1,27 +1,18 @@
-require "active_support/core_ext/time/conversions"
-
 class Mrsk::Commands::Auditor < Mrsk::Commands::Base
-  attr_reader :role
+  attr_reader :details
 
-  def initialize(config, role: nil)
+  def initialize(config, **details)
     super(config)
-    @role = role
+    @details = details
   end
 
   # Runs remotely
-  def record(line)
+  def record(line, **details)
     append \
-      [ :echo, tagged_record_line(line) ],
+      [ :echo, audit_tags(**details).except(:version, :service_version).to_s, line ],
       audit_log_file
   end
 
-  # Runs locally
-  def broadcast(line)
-    if broadcast_cmd = config.audit_broadcast_cmd
-      [ broadcast_cmd, tagged_broadcast_line(line) ]
-    end
-  end
-
   def reveal
     [ :tail, "-n", 50, audit_log_file ]
   end
@@ -31,27 +22,7 @@ class Mrsk::Commands::Auditor < Mrsk::Commands::Base
       [ "mrsk", config.service, config.destination, "audit.log" ].compact.join("-")
     end
 
-    def tagged_record_line(line)
-      tagged_line recorded_at_tag, performer_tag, role_tag, line
-    end
-
-    def tagged_broadcast_line(line)
-      tagged_line performer_tag, role_tag, line
-    end
-
-    def tagged_line(*tags_and_line)
-      "'#{tags_and_line.compact.join(" ")}'"
-    end
-
-    def recorded_at_tag
-      "[#{Time.now.to_fs(:db)}]"
-    end
-
-    def performer_tag
-      "[#{`whoami`.strip}]"
-    end
-
-    def role_tag
-      "[#{role}]" if role
+    def audit_tags(**details)
+      tags(**self.details, **details)
     end
 end

lib/mrsk/commands/base.rb

@@ -1,6 +1,9 @@
 module Mrsk::Commands
   class Base
-    delegate :redact, :argumentize, to: Mrsk::Utils
+    delegate :sensitive, :argumentize, to: Mrsk::Utils
+
+    DOCKER_HEALTH_STATUS_FORMAT = "'{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'"
+    DOCKER_HEALTH_LOG_FORMAT    = "'{{json .State.Health}}'"
 
     attr_accessor :config
 
@@ -15,8 +18,8 @@ module Mrsk::Commands
       end
     end
 
-    def container_id_for(container_name:)
-      docker :container, :ls, "--all", "--filter", "name=^#{container_name}$", "--quiet"
+    def container_id_for(container_name:, only_running: false)
+      docker :container, :ls, *("--all" unless only_running), "--filter", "name=^#{container_name}$", "--quiet"
     end
 
     private
@@ -50,5 +53,9 @@ module Mrsk::Commands
       def docker(*args)
         args.compact.unshift :docker
       end
+
+      def tags(**details)
+        Mrsk::Tags.from_config(config, **details)
+      end
   end
 end

lib/mrsk/commands/builder.rb

@@ -2,7 +2,7 @@ class Mrsk::Commands::Builder < Mrsk::Commands::Base
   delegate :create, :remove, :push, :clean, :pull, :info, to: :target
 
   def name
-    target.class.to_s.remove("Mrsk::Commands::Builder::").underscore
+    target.class.to_s.remove("Mrsk::Commands::Builder::").underscore.inquiry
   end
 
   def target
@@ -33,4 +33,24 @@ class Mrsk::Commands::Builder < Mrsk::Commands::Base
   def multiarch_remote
     @multiarch_remote ||= Mrsk::Commands::Builder::Multiarch::Remote.new(config)
   end
+
+
+  def ensure_local_dependencies_installed
+    if name.native?
+      ensure_local_docker_installed
+    else
+      combine \
+        ensure_local_docker_installed,
+        ensure_local_buildx_installed
+    end
+  end
+
+  private
+    def ensure_local_docker_installed
+      docker "--version"
+    end
+
+    def ensure_local_buildx_installed
+      docker :buildx, "version"
+    end
 end

lib/mrsk/commands/builder/base.rb

@@ -1,4 +1,7 @@
+
 class Mrsk::Commands::Builder::Base < Mrsk::Commands::Base
+  class BuilderError < StandardError; end
+
   delegate :argumentize, to: Mrsk::Utils
 
   def clean
@@ -7,7 +10,6 @@ class Mrsk::Commands::Builder::Base < Mrsk::Commands::Base
 
   def pull
     docker :pull, config.absolute_image
-    docker :pull, config.latest_image
   end
 
   def build_options
@@ -18,6 +20,7 @@ class Mrsk::Commands::Builder::Base < Mrsk::Commands::Base
     context
   end
 
+
   private
     def build_tags
       [ "-t", config.absolute_image, "-t", config.latest_image ]
@@ -28,7 +31,7 @@ class Mrsk::Commands::Builder::Base < Mrsk::Commands::Base
     end
 
     def build_args
-      argumentize "--build-arg", args, redacted: true
+      argumentize "--build-arg", args, sensitive: true
     end
 
     def build_secrets
@@ -36,7 +39,11 @@ class Mrsk::Commands::Builder::Base < Mrsk::Commands::Base
     end
 
     def build_dockerfile
+      if Pathname.new(File.expand_path(dockerfile)).exist?
         argumentize "--file", dockerfile
+      else
+        raise BuilderError, "Missing #{dockerfile}"
+      end
     end
 
     def args

lib/mrsk/commands/docker.rb

@@ -0,0 +1,21 @@
+class Mrsk::Commands::Docker < Mrsk::Commands::Base
+  # Install Docker using the https://github.com/docker/docker-install convenience script.
+  def install
+    pipe [ :curl, "-fsSL", "https://get.docker.com" ], :sh
+  end
+
+  # Checks the Docker client version. Fails if Docker is not installed.
+  def installed?
+    docker "-v"
+  end
+
+  # Checks the Docker server version. Fails if Docker is not running.
+  def running?
+    docker :version
+  end
+
+  # Do we have superuser access to install Docker and start system services?
+  def superuser?
+    [ '[ "${EUID:-$(id -u)}" -eq 0 ]' ]
+  end
+end

lib/mrsk/commands/healthcheck.rb

@@ -11,14 +11,19 @@ class Mrsk::Commands::Healthcheck < Mrsk::Commands::Base
       "--label", "service=#{container_name}",
       "-e", "MRSK_CONTAINER_NAME=\"#{container_name}\"",
       *web.env_args,
+      *web.health_check_args,
       *config.volume_args,
       *web.option_args,
       config.absolute_image,
       web.cmd
   end
 
-  def curl
-    [ :curl, "--silent", "--output", "/dev/null", "--write-out", "'%{http_code}'", "--max-time", "2", health_url ]
+  def status
+    pipe container_id, xargs(docker(:inspect, "--format", DOCKER_HEALTH_STATUS_FORMAT))
+  end
+
+  def container_health_log
+    pipe container_id, xargs(docker(:inspect, "--format", DOCKER_HEALTH_LOG_FORMAT))
   end
 
   def logs

lib/mrsk/commands/hook.rb

@@ -0,0 +1,14 @@
+class Mrsk::Commands::Hook < Mrsk::Commands::Base
+  def run(hook, **details)
+    [ hook_file(hook), env: tags(**details).env ]
+  end
+
+  def hook_exists?(hook)
+    Pathname.new(hook_file(hook)).exist?
+  end
+
+  private
+    def hook_file(hook)
+      "#{config.hooks_path}/#{hook}"
+    end
+end

lib/mrsk/commands/lock.rb

@@ -1,5 +1,5 @@
 require "active_support/duration"
-require "active_support/core_ext/numeric/time"
+require "time"
 
 class Mrsk::Commands::Lock < Mrsk::Commands::Base
   def acquire(message, version)
@@ -49,7 +49,7 @@ class Mrsk::Commands::Lock < Mrsk::Commands::Base
 
     def lock_details(message, version)
       <<~DETAILS.strip
-        Locked by: #{locked_by} at #{Time.now.gmtime}
+        Locked by: #{locked_by} at #{Time.now.utc.iso8601}
         Version: #{version}
         Message: #{message}
       DETAILS

lib/mrsk/commands/prune.rb

@@ -2,11 +2,37 @@ require "active_support/duration"
 require "active_support/core_ext/numeric/time"
 
 class Mrsk::Commands::Prune < Mrsk::Commands::Base
-  def images(until_hours: 7.days.in_hours.to_i)
-    docker :image, :prune, "--all", "--force", "--filter", "label=service=#{config.service}", "--filter", "until=#{until_hours}h"
+  def dangling_images
+    docker :image, :prune, "--force", "--filter", "label=service=#{config.service}", "--filter", "dangling=true"
   end
 
-  def containers(until_hours: 3.days.in_hours.to_i)
-    docker :container, :prune, "--force", "--filter", "label=service=#{config.service}", "--filter", "until=#{until_hours}h"
+  def tagged_images
+    pipe \
+      docker(:image, :ls, *service_filter, "--format", "'{{.ID}} {{.Repository}}:{{.Tag}}'"),
+      "grep -v -w \"#{active_image_list}\"",
+      "while read image tag; do docker rmi $tag; done"
+  end
+
+  def containers(keep_last: 5)
+    pipe \
+      docker(:ps, "-q", "-a", *service_filter, *stopped_containers_filters),
+      "tail -n +#{keep_last + 1}",
+      "while read container_id; do docker rm $container_id; done"
+  end
+
+  private
+    def stopped_containers_filters
+      [ "created", "exited", "dead" ].flat_map { |status| ["--filter", "status=#{status}"] }
+    end
+
+    def active_image_list
+      # Pull the images that are used by any containers
+      # Append repo:latest - to avoid deleting the latest tag
+      # Append repo:<none> - to avoid deleting dangling images that are in use. Unused dangling images are deleted separately
+      "$(docker container ls -a --format '{{.Image}}\\|' --filter label=service=#{config.service} | tr -d '\\n')#{config.latest_image}\\|#{config.repository}:<none>"
+    end
+
+    def service_filter
+      [ "--filter", "label=service=#{config.service}" ]
     end
 end

lib/mrsk/commands/registry.rb

@@ -2,7 +2,7 @@ class Mrsk::Commands::Registry < Mrsk::Commands::Base
   delegate :registry, to: :config
 
   def login
-    docker :login, registry["server"], "-u", redact(lookup("username")), "-p", redact(lookup("password"))
+    docker :login, registry["server"], "-u", sensitive(lookup("username")), "-p", sensitive(lookup("password"))
   end
 
   def logout

lib/mrsk/commands/traefik.rb

@@ -1,7 +1,7 @@
 class Mrsk::Commands::Traefik < Mrsk::Commands::Base
-  delegate :optionize, to: Mrsk::Utils
+  delegate :argumentize, :argumentize_env_with_secrets, :optionize, to: Mrsk::Utils
 
-  IMAGE = "traefik:v2.9.9"
+  DEFAULT_IMAGE = "traefik:v2.9"
   CONTAINER_PORT = 80
 
   def run
@@ -10,9 +10,11 @@ class Mrsk::Commands::Traefik < Mrsk::Commands::Base
       "--restart", "unless-stopped",
       "--publish", port,
       "--volume", "/var/run/docker.sock:/var/run/docker.sock",
+      *env_args,
       *config.logging_args,
+      *label_args,
       *docker_options_args,
-      IMAGE,
+      image,
       "--providers.docker",
       "--log.level=DEBUG",
       *cmd_option_args
@@ -56,6 +58,28 @@ class Mrsk::Commands::Traefik < Mrsk::Commands::Base
   end
 
   private
+    def label_args
+      argumentize "--label", labels
+    end
+
+    def env_args
+      env_config = config.traefik["env"] || {}
+
+      if env_config.present?
+        argumentize_env_with_secrets(env_config)
+      else
+        []
+      end
+    end
+
+    def labels
+      config.traefik["labels"] || []
+    end
+
+    def image
+      config.traefik.fetch("image") { DEFAULT_IMAGE }
+    end
+
     def docker_options_args
       optionize(config.traefik["options"] || {})
     end

lib/mrsk/configuration.rb

@@ -6,7 +6,7 @@ require "erb"
 require "net/ssh/proxy/jump"
 
 class Mrsk::Configuration
-  delegate :service, :image, :servers, :env, :labels, :registry, :builder, :stop_wait_time, to: :raw_config, allow_nil: true
+  delegate :service, :image, :servers, :env, :labels, :registry, :builder, :stop_wait_time, :hooks_path, to: :raw_config, allow_nil: true
   delegate :argumentize, :argumentize_env_with_secrets, :optionize, to: Mrsk::Utils
 
   attr_accessor :destination
@@ -50,7 +50,7 @@ class Mrsk::Configuration
   end
 
   def version
-    @declared_version.presence || ENV["VERSION"] || current_commit_hash
+    @declared_version.presence || ENV["VERSION"] || git_version
   end
 
   def abbreviated_version
@@ -87,6 +87,10 @@ class Mrsk::Configuration
     roles.select(&:running_traefik?).flat_map(&:hosts).uniq
   end
 
+  def boot
+    Mrsk::Configuration::Boot.new(config: self)
+  end
+
 
   def repository
     [ raw_config.registry["server"], image ].compact.join("/")
@@ -143,6 +147,8 @@ class Mrsk::Configuration
     if raw_config.ssh.present? && raw_config.ssh["proxy"]
       Net::SSH::Proxy::Jump.new \
         raw_config.ssh["proxy"].include?("@") ? raw_config.ssh["proxy"] : "root@#{raw_config.ssh["proxy"]}"
+    elsif raw_config.ssh.present? && raw_config.ssh["proxy_command"]
+      Net::SSH::Proxy::Command.new(raw_config.ssh["proxy_command"])
     end
   end
 
@@ -151,12 +157,8 @@ class Mrsk::Configuration
   end
 
 
-  def audit_broadcast_cmd
-    raw_config.audit_broadcast_cmd
-  end
-
   def healthcheck
-    { "path" => "/up", "port" => 3000 }.merge(raw_config.healthcheck || {})
+    { "path" => "/up", "port" => 3000, "max_attempts" => 7 }.merge(raw_config.healthcheck || {})
   end
 
   def readiness_delay
@@ -191,6 +193,10 @@ class Mrsk::Configuration
     raw_config.traefik || {}
   end
 
+  def hooks_path
+    raw_config.hooks_path || ".mrsk/hooks"
+  end
+
   private
     # Will raise ArgumentError if any required config keys are missing
     def ensure_required_keys_present
@@ -227,10 +233,12 @@ class Mrsk::Configuration
       raw_config.servers.is_a?(Array) ? [ "web" ] : raw_config.servers.keys.sort
     end
 
-    def current_commit_hash
-      @current_commit_hash ||=
+    def git_version
+      @git_version ||=
         if system("git rev-parse")
-          `git rev-parse HEAD`.strip
+          uncommitted_suffix = `git status --porcelain`.strip.present? ? "_uncommitted_#{SecureRandom.hex(8)}" : ""
+
+          "#{`git rev-parse HEAD`.strip}#{uncommitted_suffix}"
         else
           raise "Can't use commit hash as version, no git repository found in #{Dir.pwd}"
         end

lib/mrsk/configuration/boot.rb

@@ -0,0 +1,20 @@
+class Mrsk::Configuration::Boot
+  def initialize(config:)
+    @options = config.raw_config.boot || {}
+    @host_count = config.all_hosts.count
+  end
+
+  def limit
+    limit = @options["limit"]
+
+    if limit.to_s.end_with?("%")
+      @host_count * limit.to_i / 100
+    else
+      limit
+    end
+  end
+
+  def wait
+    @options["wait"]
+  end
+end

lib/mrsk/configuration/role.rb

@@ -35,6 +35,28 @@ class Mrsk::Configuration::Role
     argumentize_env_with_secrets env
   end
 
+  def health_check_args
+    if health_check_cmd.present?
+      optionize({ "health-cmd" => health_check_cmd, "health-interval" => health_check_interval })
+    else
+      []
+    end
+  end
+
+  def health_check_cmd
+    options = specializations["healthcheck"] || {}
+    options = config.healthcheck.merge(options) if running_traefik?
+
+    options["cmd"] || http_health_check(port: options["port"], path: options["path"])
+  end
+
+  def health_check_interval
+    options = specializations["healthcheck"] || {}
+    options = config.healthcheck.merge(options) if running_traefik?
+
+    options["interval"] || "1s"
+  end
+
   def cmd
     specializations["cmd"]
   end
@@ -74,18 +96,23 @@ class Mrsk::Configuration::Role
     def traefik_labels
       if running_traefik?
         {
-          "traefik.http.routers.#{config.service}.rule" => "PathPrefix(`/`)",
-          "traefik.http.services.#{config.service}.loadbalancer.healthcheck.path" => config.healthcheck["path"],
-          "traefik.http.services.#{config.service}.loadbalancer.healthcheck.interval" => "1s",
-          "traefik.http.middlewares.#{config.service}-retry.retry.attempts" => "5",
-          "traefik.http.middlewares.#{config.service}-retry.retry.initialinterval" => "500ms",
-          "traefik.http.routers.#{config.service}.middlewares" => "#{config.service}-retry@docker"
+          # Setting a service property ensures that the generated service name will be consistent between versions
+          "traefik.http.services.#{traefik_service}.loadbalancer.server.scheme" => "http",
+
+          "traefik.http.routers.#{traefik_service}.rule" => "PathPrefix(`/`)",
+          "traefik.http.middlewares.#{traefik_service}-retry.retry.attempts" => "5",
+          "traefik.http.middlewares.#{traefik_service}-retry.retry.initialinterval" => "500ms",
+          "traefik.http.routers.#{traefik_service}.middlewares" => "#{traefik_service}-retry@docker"
         }
       else
         {}
       end
     end
 
+    def traefik_service
+      [ config.service, name, config.destination ].compact.join("-")
+    end
+
     def custom_labels
       Hash.new.tap do |labels|
         labels.merge!(config.labels) if config.labels.present?
@@ -121,4 +148,8 @@ class Mrsk::Configuration::Role
         new_env["clear"] = (clear_app_env + clear_role_env).uniq
       end
     end
+
+    def http_health_check(port:, path:)
+      "curl -f #{URI.join("http://localhost:#{port}", path)} || exit 1" if path.present? || port.present?
+    end
 end

lib/mrsk/sshkit_with_ext.rb

@@ -1,12 +1,56 @@
 require "sshkit"
 require "sshkit/dsl"
+require "active_support/core_ext/hash/deep_merge"
+require "json"
 
 class SSHKit::Backend::Abstract
-  def capture_with_info(*args)
-    capture(*args, verbosity: Logger::INFO)
+  def capture_with_info(*args, **kwargs)
+    capture(*args, **kwargs, verbosity: Logger::INFO)
+  end
+
+  def capture_with_debug(*args, **kwargs)
+    capture(*args, **kwargs, verbosity: Logger::DEBUG)
+  end
+
+  def capture_with_pretty_json(*args, **kwargs)
+    JSON.pretty_generate(JSON.parse(capture(*args, **kwargs)))
   end
 
   def puts_by_host(host, output, type: "App")
     puts "#{type} Host: #{host}\n#{output}\n\n"
   end
+
+  # Our execution pattern is for the CLI execute args lists returned
+  # from commands, but this doesn't support returning execution options
+  # from the command.
+  #
+  # Support this by using kwargs for CLI options and merging with the
+  # args-extracted options.
+  module CommandEnvMerge
+    private
+
+    # Override to merge options returned by commands in the args list with
+    # options passed by the CLI and pass them along as kwargs.
+    def command(args, options)
+      more_options, args = args.partition { |a| a.is_a? Hash }
+      more_options << options
+
+      build_command(args, **more_options.reduce(:deep_merge))
+    end
+
+    # Destructure options to pluck out env for merge
+    def build_command(args, env: nil, **options)
+      # Rely on native Ruby kwargs precedence rather than explicit Hash merges
+      SSHKit::Command.new(*args, **default_command_options, **options, env: env_for(env))
+    end
+
+    def default_command_options
+      { in: pwd_path, host: @host, user: @user, group: @group }
+    end
+
+    def env_for(env)
+      @env.to_h.merge(env.to_h)
+    end
+  end
+  prepend CommandEnvMerge
 end

lib/mrsk/tags.rb

@@ -0,0 +1,39 @@
+require "time"
+
+class Mrsk::Tags
+  attr_reader :config, :tags
+
+  class << self
+    def from_config(config, **extra)
+      new(**default_tags(config), **extra)
+    end
+
+    def default_tags(config)
+      { recorded_at: Time.now.utc.iso8601,
+        performer: `whoami`.chomp,
+        destination: config.destination,
+        version: config.version,
+        service_version: service_version(config) }
+    end
+
+    def service_version(config)
+      [ config.service, config.abbreviated_version ].compact.join("@")
+    end
+  end
+
+  def initialize(**tags)
+    @tags = tags.compact
+  end
+
+  def env
+    tags.transform_keys { |detail| "MRSK_#{detail.upcase}" }
+  end
+
+  def to_s
+    tags.values.map { |value| "[#{value}]" }.join(" ")
+  end
+
+  def except(*tags)
+    self.class.new(**self.tags.except(*tags))
+  end
+end

lib/mrsk/utils.rb

@@ -1,12 +1,15 @@
 module Mrsk::Utils
   extend self
 
+  DOLLAR_SIGN_WITHOUT_SHELL_EXPANSION_REGEX = /\$(?!{[^\}]*\})/
+
   # Return a list of escaped shell arguments using the same named argument against the passed attributes (hash or array).
-  def argumentize(argument, attributes, redacted: false)
+  def argumentize(argument, attributes, sensitive: false)
     Array(attributes).flat_map do |key, value|
       if value.present?
-        escaped_pair = [ key, escape_shell_value(value) ].join("=")
-        [ argument, redacted ? redact(escaped_pair) : escaped_pair ]
+        attr = "#{key}=#{escape_shell_value(value)}"
+        attr = self.sensitive(attr, redaction: "#{key}=[REDACTED]") if sensitive
+        [ argument, attr]
       else
         [ argument, key ]
       end
@@ -17,7 +20,7 @@ module Mrsk::Utils
   # but redacts and expands secrets.
   def argumentize_env_with_secrets(env)
     if (secrets = env["secret"]).present?
-      argumentize("-e", secrets.to_h { |key| [ key, ENV.fetch(key) ] }, redacted: true) + argumentize("-e", env["clear"])
+      argumentize("-e", secrets.to_h { |key| [ key, ENV.fetch(key) ] }, sensitive: true) + argumentize("-e", env["clear"])
     else
       argumentize "-e", env.fetch("clear", env)
     end
@@ -39,18 +42,55 @@ module Mrsk::Utils
     args.flat_map { |key, value| value.try(:map) { |entry| [key, entry] } || [ [ key, value ] ] }
   end
 
-  # Copied from SSHKit::Backend::Abstract#redact to be available inside Commands classes
-  def redact(arg) # Used in execute_command to hide redact() args a user passes in
-    arg.to_s.extend(SSHKit::Redaction) # to_s due to our inability to extend Integer, etc
+  # Marks sensitive values for redaction in logs and human-visible output.
+  # Pass `redaction:` to change the default `"[REDACTED]"` redaction, e.g.
+  # `sensitive "#{arg}=#{secret}", redaction: "#{arg}=xxxx"
+  def sensitive(...)
+    Mrsk::Utils::Sensitive.new(...)
+  end
+
+  def redacted(value)
+    case
+    when value.respond_to?(:redaction)
+      value.redaction
+    when value.respond_to?(:transform_values)
+      value.transform_values { |value| redacted value }
+    when value.respond_to?(:map)
+      value.map { |element| redacted element }
+    else
+      value
+    end
+  end
+
+  def unredacted(value)
+    case
+    when value.respond_to?(:unredacted)
+      value.unredacted
+    when value.respond_to?(:transform_values)
+      value.transform_values { |value| unredacted value }
+    when value.respond_to?(:map)
+      value.map { |element| unredacted element }
+    else
+      value
+    end
   end
 
   # Escape a value to make it safe for shell use.
   def escape_shell_value(value)
-    value.to_s.dump.gsub(/`/, '\\\\`')
+    value.to_s.dump
+      .gsub(/`/, '\\\\`')
+      .gsub(DOLLAR_SIGN_WITHOUT_SHELL_EXPANSION_REGEX, '\$')
   end
 
   # Abbreviate a git revhash for concise display
   def abbreviate_version(version)
-    version[0...7] if version
+    if version
+      # Don't abbreviate <sha>_uncommitted_<etc>
+      if version.include?("_")
+        version
+      else
+        version[0...7]
+      end
+    end
   end
 end

lib/mrsk/utils/healthcheck_poller.rb

@@ -0,0 +1,39 @@
+class Mrsk::Utils::HealthcheckPoller
+  TRAEFIK_HEALTHY_DELAY = 2
+
+  class HealthcheckError < StandardError; end
+
+  class << self
+    def wait_for_healthy(pause_after_ready: false, &block)
+      attempt = 1
+      max_attempts = MRSK.config.healthcheck["max_attempts"]
+
+      begin
+        case status = block.call
+        when "healthy"
+          sleep TRAEFIK_HEALTHY_DELAY if pause_after_ready
+        when "running" # No health check configured
+          sleep MRSK.config.readiness_delay if pause_after_ready
+        else
+          raise HealthcheckError, "container not ready (#{status})"
+        end
+      rescue HealthcheckError => e
+        if attempt <= max_attempts
+          info "#{e.message}, retrying in #{attempt}s (attempt #{attempt}/#{max_attempts})..."
+          sleep attempt
+          attempt += 1
+          retry
+        else
+          raise
+        end
+      end
+
+      info "Container is healthy!"
+    end
+
+    private
+      def info(message)
+        SSHKit.config.output.info(message)
+      end
+  end
+end

lib/mrsk/utils/sensitive.rb

@@ -0,0 +1,19 @@
+require "active_support/core_ext/module/delegation"
+
+class Mrsk::Utils::Sensitive
+  # So SSHKit knows to redact these values.
+  include SSHKit::Redaction
+
+  attr_reader :unredacted, :redaction
+  delegate :to_s, to: :unredacted
+  delegate :inspect, to: :redaction
+
+  def initialize(value, redaction: "[REDACTED]")
+    @unredacted, @redaction = value, redaction
+  end
+
+  # Sensitive values won't leak into YAML output.
+  def encode_with(coder)
+    coder.represent_scalar nil, redaction
+  end
+end

lib/mrsk/version.rb

@@ -1,3 +1,3 @@
 module Mrsk
-  VERSION = "0.10.0"
+  VERSION = "0.13.2"
 end

mrsk.gemspec

@@ -14,6 +14,7 @@ Gem::Specification.new do |spec|
 
   spec.add_dependency "activesupport", ">= 7.0"
   spec.add_dependency "sshkit", "~> 1.21"
+  spec.add_dependency "net-ssh", "~> 7.0"
   spec.add_dependency "thor", "~> 1.2"
   spec.add_dependency "dotenv", "~> 2.8"
   spec.add_dependency "zeitwerk", "~> 2.5"

test/cli/app_test.rb

@@ -2,11 +2,10 @@ require_relative "cli_test_case"
 
 class CliAppTest < CliTestCase
   test "boot" do
-    # Stub current version fetch
-    SSHKit::Backend::Abstract.any_instance.stubs(:capture).returns("123") # old version
-
+    stub_running
     run_command("boot").tap do |output|
-      assert_match "docker run --detach --restart unless-stopped", output
+      assert_match "docker tag dhh/app:latest dhh/app:latest", output
+      assert_match /docker run --detach --restart unless-stopped --name app-web-latest --hostname 1.1.1.1-[0-9a-f]{12} /, output
       assert_match "docker container ls --all --filter name=^app-web-123$ --quiet | xargs docker stop", output
     end
   end
@@ -15,23 +14,49 @@ class CliAppTest < CliTestCase
     run_command("details") # Preheat MRSK const
 
     SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
-      .with(:docker, :container, :ls, "--all", "--filter", "name=^app-web-latest$", "--quiet")
+      .with(:docker, :container, :ls, "--filter", "name=^app-web-latest$", "--quiet", raise_on_non_zero_exit: false)
       .returns("12345678") # running version
 
     SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
-      .with(:docker, :ps, "--filter", "label=service=app", "--filter", "label=role=web", "--format", "\"{{.Names}}\"", "|", "sed 's/-/\\n/g'", "|", "tail -n 1")
+      .with(:docker, :container, :ls, "--all", "--filter", "name=^app-web-latest$", "--quiet", "|", :xargs, :docker, :inspect, "--format", "'{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'")
+      .returns("running") # health check
+
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:docker, :ps, "--filter", "label=service=app", "--filter", "label=role=web", "--filter", "status=running", "--filter", "status=restarting", "--latest", "--format", "\"{{.Names}}\"", "|", "grep -oE \"\\-[^-]+$\"", "|", "cut -c 2-", raise_on_non_zero_exit: false)
       .returns("123") # old version
 
     run_command("boot").tap do |output|
       assert_match /Renaming container .* to .* as already deployed on 1.1.1.1/, output # Rename
-      assert_match /docker rename .* .*/, output
-      assert_match "docker run --detach --restart unless-stopped", output
+      assert_match /docker rename app-web-latest app-web-latest_replaced_[0-9a-f]{16}/, output
+      assert_match /docker run --detach --restart unless-stopped --name app-web-latest --hostname 1.1.1.1-[0-9a-f]{12} /, output
       assert_match "docker container ls --all --filter name=^app-web-123$ --quiet | xargs docker stop", output
     end
   ensure
     Thread.report_on_exception = true
   end
 
+  test "boot uses group strategy when specified" do
+    Mrsk::Cli::App.any_instance.stubs(:on).with("1.1.1.1").twice # acquire & release lock
+    Mrsk::Cli::App.any_instance.stubs(:on).with([ "1.1.1.1" ]) # tag container
+
+    # Strategy is used when booting the containers
+    Mrsk::Cli::App.any_instance.expects(:on).with([ "1.1.1.1" ], in: :groups, limit: 3, wait: 2).with_block_given
+
+    run_command("boot", config: :with_boot_strategy)
+  end
+
+  test "boot errors leave lock in place" do
+    invoke_options = { "config_file" => "test/fixtures/deploy_simple.yml", "version" => "999" }
+
+    Mrsk::Cli::App.any_instance.expects(:using_version).raises(RuntimeError)
+
+    assert !MRSK.holding_lock?
+    assert_raises(RuntimeError) do
+      stderred { run_command("boot") }
+    end
+    assert MRSK.holding_lock?
+  end
+
   test "start" do
     run_command("start").tap do |output|
       assert_match "docker start app-web-999", output
@@ -40,7 +65,28 @@ class CliAppTest < CliTestCase
 
   test "stop" do
     run_command("stop").tap do |output|
-      assert_match "docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker stop", output
+      assert_match "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker stop", output
+    end
+  end
+
+  test "stale_containers" do
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:docker, :ps, "--filter", "label=service=app", "--filter", "label=role=web", "--format", "\"{{.Names}}\"", "|", "grep -oE \"\\-[^-]+$\"", "|", "cut -c 2-", raise_on_non_zero_exit: false)
+      .returns("12345678\n87654321")
+
+    run_command("stale_containers").tap do |output|
+      assert_match /Detected stale container for role web with version 87654321/, output
+    end
+  end
+
+  test "stop stale_containers" do
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:docker, :ps, "--filter", "label=service=app", "--filter", "label=role=web", "--format", "\"{{.Names}}\"", "|", "grep -oE \"\\-[^-]+$\"", "|", "cut -c 2-", raise_on_non_zero_exit: false)
+      .returns("12345678\n87654321")
+
+    run_command("stale_containers", "--stop").tap do |output|
+      assert_match /Stopping stale container for role web with version 87654321/, output
+      assert_match /#{Regexp.escape("docker container ls --all --filter name=^app-web-87654321$ --quiet | xargs docker stop")}/, output
     end
   end
 
@@ -52,7 +98,7 @@ class CliAppTest < CliTestCase
 
   test "remove" do
     run_command("remove").tap do |output|
-      assert_match /#{Regexp.escape("docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker stop")}/, output
+      assert_match /#{Regexp.escape("docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker stop")}/, output
       assert_match /#{Regexp.escape("docker container prune --force --filter label=service=app")}/, output
       assert_match /#{Regexp.escape("docker image prune --all --force --filter label=service=app")}/, output
     end
@@ -84,7 +130,7 @@ class CliAppTest < CliTestCase
 
   test "exec with reuse" do
     run_command("exec", "--reuse", "ruby -v").tap do |output|
-      assert_match "docker ps --filter label=service=app --format \"{{.Names}}\" | sed 's/-/\\n/g' | tail -n 1", output # Get current version
+      assert_match "docker ps --filter label=service=app --filter status=running --filter status=restarting --latest --format \"{{.Names}}\" | grep -oE \"\\-[^-]+$\" | cut -c 2-", output # Get current version
       assert_match "docker exec app-web-999 ruby -v", output
     end
   end
@@ -103,33 +149,41 @@ class CliAppTest < CliTestCase
 
   test "logs" do
     SSHKit::Backend::Abstract.any_instance.stubs(:exec)
-      .with("ssh -t root@1.1.1.1 'docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker logs --timestamps --tail 10 2>&1'")
+      .with("ssh -t root@1.1.1.1 'docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest| xargs docker logs --timestamps --tail 10 2>&1'")
 
-    assert_match "docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker logs --tail 100 2>&1", run_command("logs")
+    assert_match "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs --tail 100 2>&1", run_command("logs")
   end
 
   test "logs with follow" do
     SSHKit::Backend::Abstract.any_instance.stubs(:exec)
-      .with("ssh -t root@1.1.1.1 'docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker logs --timestamps --tail 10 --follow 2>&1'")
+      .with("ssh -t root@1.1.1.1 'docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs --timestamps --tail 10 --follow 2>&1'")
 
-    assert_match "docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker logs --timestamps --tail 10 --follow 2>&1", run_command("logs", "--follow")
+    assert_match "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs --timestamps --tail 10 --follow 2>&1", run_command("logs", "--follow")
   end
 
   test "version" do
     run_command("version").tap do |output|
-      assert_match "docker ps --filter label=service=app --format \"{{.Names}}\" | sed 's/-/\\n/g' | tail -n 1", output
+      assert_match "docker ps --filter label=service=app --filter status=running --filter status=restarting --latest --format \"{{.Names}}\" | grep -oE \"\\-[^-]+$\" | cut -c 2-", output
     end
   end
 
 
   test "version through main" do
     stdouted { Mrsk::Cli::Main.start(["app", "version", "-c", "test/fixtures/deploy_with_accessories.yml", "--hosts", "1.1.1.1"]) }.tap do |output|
-      assert_match "docker ps --filter label=service=app --format \"{{.Names}}\" | sed 's/-/\\n/g' | tail -n 1", output
+      assert_match "docker ps --filter label=service=app --filter status=running --filter status=restarting --latest --format \"{{.Names}}\" | grep -oE \"\\-[^-]+$\" | cut -c 2-", output
     end
   end
 
   private
-    def run_command(*command)
-      stdouted { Mrsk::Cli::App.start([*command, "-c", "test/fixtures/deploy_with_accessories.yml", "--hosts", "1.1.1.1"]) }
+    def run_command(*command, config: :with_accessories)
+      stdouted { Mrsk::Cli::App.start([*command, "-c", "test/fixtures/deploy_#{config}.yml", "--hosts", "1.1.1.1"]) }
+    end
+
+    def stub_running
+      SSHKit::Backend::Abstract.any_instance.stubs(:capture_with_info).returns("123") # old version
+
+      SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+        .with(:docker, :container, :ls, "--all", "--filter", "name=^app-web-latest$", "--quiet", "|", :xargs, :docker, :inspect, "--format", "'{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'")
+        .returns("running") # health check
     end
 end

test/cli/build_test.rb

@@ -9,7 +9,12 @@ class CliBuildTest < CliTestCase
   end
 
   test "push" do
+    Mrsk::Commands::Hook.any_instance.stubs(:hook_exists?).returns(true)
+    hook_variables = { version: 999, service_version: "app@999", hosts: "1.1.1.1,1.1.1.2,1.1.1.3,1.1.1.4", command: "build", subcommand: "push" }
+
     run_command("push").tap do |output|
+      assert_hook_ran "pre-build", output, **hook_variables
+      assert_match /docker --version && docker buildx version/, output
       assert_match /docker buildx build --push --platform linux\/amd64,linux\/arm64 --builder mrsk-app-multiarch -t dhh\/app:999 -t dhh\/app:latest --label service="app" --file Dockerfile \. as .*@localhost/, output
     end
   end
@@ -17,7 +22,10 @@ class CliBuildTest < CliTestCase
   test "push without builder" do
     stub_locking
     SSHKit::Backend::Abstract.any_instance.stubs(:execute)
-      .with { |arg| arg == :docker }
+      .with(:docker, "--version", "&&", :docker, :buildx, "version")
+
+    SSHKit::Backend::Abstract.any_instance.stubs(:execute)
+      .with { |*args| args[0..1] == [:docker, :buildx] }
       .raises(SSHKit::Command::Failed.new("no builder"))
       .then
       .returns(true)
@@ -27,10 +35,28 @@ class CliBuildTest < CliTestCase
     end
   end
 
+  test "push with no buildx plugin" do
+    stub_locking
+    SSHKit::Backend::Abstract.any_instance.stubs(:execute)
+      .with(:docker, "--version", "&&", :docker, :buildx, "version")
+      .raises(SSHKit::Command::Failed.new("no buildx"))
+
+    Mrsk::Commands::Builder.any_instance.stubs(:native_and_local?).returns(false)
+    assert_raises(Mrsk::Cli::Build::BuildError) { run_command("push") }
+  end
+
+  test "push pre-build hook failure" do
+    fail_hook("pre-build")
+
+    assert_raises(Mrsk::Cli::HookError) { run_command("push") }
+
+    assert @executions.none? { |args| args[0..2] == [:docker, :buildx, :build] }
+  end
+
   test "pull" do
     run_command("pull").tap do |output|
       assert_match /docker image rm --force dhh\/app:999/, output
-      assert_match /docker pull dhh\/app:latest/, output
+      assert_match /docker pull dhh\/app:999/, output
     end
   end
 
@@ -73,10 +99,10 @@ class CliBuildTest < CliTestCase
       stdouted { Mrsk::Cli::Build.start([*command, "-c", "test/fixtures/deploy_with_accessories.yml"]) }
     end
 
-    def stub_locking
+    def stub_dependency_checks
       SSHKit::Backend::Abstract.any_instance.stubs(:execute)
-        .with { |arg1, arg2| arg1 == :mkdir && arg2 == :mrsk_lock }
+        .with(:docker, "--version", "&&", :docker, :buildx, "version")
       SSHKit::Backend::Abstract.any_instance.stubs(:execute)
-        .with { |arg1, arg2| arg1 == :rm && arg2 == "mrsk_lock/details" }
+        .with { |*args| args[0..1] == [:docker, :buildx] }
     end
 end

test/cli/cli_test_case.rb

@@ -1,9 +1,6 @@
 require "test_helper"
-require "active_support/testing/stream"
 
 class CliTestCase < ActiveSupport::TestCase
-  include ActiveSupport::Testing::Stream
-
   setup do
     ENV["VERSION"]             = "999"
     ENV["RAILS_MASTER_KEY"]    = "123"
@@ -19,7 +16,41 @@ class CliTestCase < ActiveSupport::TestCase
   end
 
   private
-    def stdouted
-      capture(:stdout) { yield }.strip
+    def fail_hook(hook)
+      @executions = []
+      Mrsk::Commands::Hook.any_instance.stubs(:hook_exists?).returns(true)
+
+      SSHKit::Backend::Abstract.any_instance.stubs(:execute)
+        .with { |*args| @executions << args; args != [".mrsk/hooks/#{hook}"] }
+      SSHKit::Backend::Abstract.any_instance.stubs(:execute)
+        .with { |*args| args.first == ".mrsk/hooks/#{hook}" }
+        .raises(SSHKit::Command::Failed.new("failed"))
+    end
+
+    def stub_locking
+      SSHKit::Backend::Abstract.any_instance.stubs(:execute)
+        .with { |arg1, arg2| arg1 == :mkdir && arg2 == :mrsk_lock }
+      SSHKit::Backend::Abstract.any_instance.stubs(:execute)
+        .with { |arg1, arg2| arg1 == :rm && arg2 == "mrsk_lock/details" }
+    end
+
+    def assert_hook_ran(hook, output, version:, service_version:, hosts:, command:, subcommand: nil, runtime: nil)
+      performer = `whoami`.strip
+
+      assert_match "Running the #{hook} hook...\n", output
+
+      expected = %r{Running\s/usr/bin/env\s\.mrsk/hooks/#{hook}\sas\s#{performer}@localhost\n\s
+        DEBUG\s\[[0-9a-f]*\]\sCommand:\s\(\sexport\s
+        MRSK_RECORDED_AT=\"\d\d\d\d-\d\d-\d\dT\d\d:\d\d:\d\dZ\"\s
+        MRSK_PERFORMER=\"#{performer}\"\s
+        MRSK_VERSION=\"#{version}\"\s
+        MRSK_SERVICE_VERSION=\"#{service_version}\"\s
+        MRSK_HOSTS=\"#{hosts}\"\s
+        MRSK_COMMAND=\"#{command}\"\s
+        #{"MRSK_SUBCOMMAND=\\\"#{subcommand}\\\"\\s" if subcommand}
+        #{"MRSK_RUNTIME=\\\"#{runtime}\\\"\\s" if runtime}
+        ;\s/usr/bin/env\s\.mrsk/hooks/#{hook} }x
+
+      assert_match expected, output
     end
 end

test/cli/healthcheck_test.rb

@@ -5,62 +5,63 @@ class CliHealthcheckTest < CliTestCase
     # Prevent expected failures from outputting to terminal
     Thread.report_on_exception = false
 
-    SSHKit::Backend::Abstract.any_instance.stubs(:sleep) # No sleeping when retrying
+    Mrsk::Utils::HealthcheckPoller.stubs(:sleep) # No sleeping when retrying
+
     SSHKit::Backend::Abstract.any_instance.stubs(:execute)
       .with(:docker, :container, :ls, "--all", "--filter", "name=^healthcheck-app-999$", "--quiet", "|", :xargs, :docker, :stop, raise_on_non_zero_exit: false)
     SSHKit::Backend::Abstract.any_instance.stubs(:execute)
-      .with(:docker, :run, "--detach", "--name", "healthcheck-app-999", "--publish", "3999:3000", "--label", "service=healthcheck-app", "-e", "MRSK_CONTAINER_NAME=\"healthcheck-app\"", "dhh/app:999")
+      .with(:docker, :run, "--detach", "--name", "healthcheck-app-999", "--publish", "3999:3000", "--label", "service=healthcheck-app", "-e", "MRSK_CONTAINER_NAME=\"healthcheck-app\"", "--health-cmd", "\"curl -f http://localhost:3000/up || exit 1\"", "--health-interval", "\"1s\"", "dhh/app:999")
     SSHKit::Backend::Abstract.any_instance.stubs(:execute)
       .with(:docker, :container, :ls, "--all", "--filter", "name=^healthcheck-app-999$", "--quiet", "|", :xargs, :docker, :container, :rm, raise_on_non_zero_exit: false)
 
     # Fail twice to test retry logic
     SSHKit::Backend::Abstract.any_instance.stubs(:capture_with_info)
-      .with(:curl, "--silent", "--output", "/dev/null", "--write-out", "'%{http_code}'", "--max-time", "2", "http://localhost:3999/up")
-      .raises(SSHKit::Command::Failed)
+      .with(:docker, :container, :ls, "--all", "--filter", "name=^healthcheck-app-999$", "--quiet", "|", :xargs, :docker, :inspect, "--format", "'{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'")
+      .returns("starting")
       .then
-      .raises(SSHKit::Command::Failed)
+      .returns("unhealthy")
       .then
-      .returns("200")
+      .returns("healthy")
 
     run_command("perform").tap do |output|
-      assert_match "Health check against /up failed to respond, retrying in 1s...", output
-      assert_match "Health check against /up failed to respond, retrying in 2s...", output
-      assert_match "Health check against /up succeeded with 200 OK!", output
+      assert_match "container not ready (starting), retrying in 1s (attempt 1/7)...", output
+      assert_match "container not ready (unhealthy), retrying in 2s (attempt 2/7)...", output
+      assert_match "Container is healthy!", output
     end
   end
 
-  test "perform failing because of curl" do
+  test "perform failing to become healthy" do
     # Prevent expected failures from outputting to terminal
     Thread.report_on_exception = false
 
-    SSHKit::Backend::Abstract.any_instance.stubs(:execute) # No need to execute anything here
+    Mrsk::Utils::HealthcheckPoller.stubs(:sleep) # No sleeping when retrying
+
+    SSHKit::Backend::Abstract.any_instance.stubs(:execute)
+      .with(:docker, :container, :ls, "--all", "--filter", "name=^healthcheck-app-999$", "--quiet", "|", :xargs, :docker, :stop, raise_on_non_zero_exit: false)
+    SSHKit::Backend::Abstract.any_instance.stubs(:execute)
+      .with(:docker, :run, "--detach", "--name", "healthcheck-app-999", "--publish", "3999:3000", "--label", "service=healthcheck-app", "-e", "MRSK_CONTAINER_NAME=\"healthcheck-app\"", "--health-cmd", "\"curl -f http://localhost:3000/up || exit 1\"", "--health-interval", "\"1s\"", "dhh/app:999")
+    SSHKit::Backend::Abstract.any_instance.stubs(:execute)
+      .with(:docker, :container, :ls, "--all", "--filter", "name=^healthcheck-app-999$", "--quiet", "|", :xargs, :docker, :container, :rm, raise_on_non_zero_exit: false)
+
+    # Continually report unhealthy
     SSHKit::Backend::Abstract.any_instance.stubs(:capture_with_info)
-      .with(:curl, "--silent", "--output", "/dev/null", "--write-out", "'%{http_code}'", "--max-time", "2", "http://localhost:3999/up")
-      .returns("curl: command not found")
+      .with(:docker, :container, :ls, "--all", "--filter", "name=^healthcheck-app-999$", "--quiet", "|", :xargs, :docker, :inspect, "--format", "'{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'")
+      .returns("unhealthy")
+
+    # Capture logs when failing
     SSHKit::Backend::Abstract.any_instance.stubs(:capture_with_info)
       .with(:docker, :container, :ls, "--all", "--filter", "name=^healthcheck-app-999$", "--quiet", "|", :xargs, :docker, :logs, "--tail", 50, "2>&1")
+      .returns("some log output")
 
-    exception = assert_raises SSHKit::Runner::ExecuteError do
-      run_command("perform")
-    end
-    assert_match "Health check against /up failed to return 200 OK!", exception.message
-  end
-
-  test "perform failing for unknown reason" do
-    # Prevent expected failures from outputting to terminal
-    Thread.report_on_exception = false
-
-    SSHKit::Backend::Abstract.any_instance.stubs(:execute) # No need to execute anything here
-    SSHKit::Backend::Abstract.any_instance.stubs(:capture_with_info)
-      .with(:curl, "--silent", "--output", "/dev/null", "--write-out", "'%{http_code}'", "--max-time", "2", "http://localhost:3999/up")
-      .returns("500")
-    SSHKit::Backend::Abstract.any_instance.stubs(:capture_with_info)
-      .with(:docker, :container, :ls, "--all", "--filter", "name=^healthcheck-app-999$", "--quiet", "|", :xargs, :docker, :logs, "--tail", 50, "2>&1")
+    # Capture container health log when failing
+    SSHKit::Backend::Abstract.any_instance.stubs(:capture_with_pretty_json)
+      .with(:docker, :container, :ls, "--all", "--filter", "name=^healthcheck-app-999$", "--quiet", "|", :xargs, :docker, :inspect, "--format",  "'{{json .State.Health}}'")
+      .returns('{"Status":"unhealthy","Log":[{"ExitCode": 1,"Output": "/bin/sh: 1: curl: not found\n"}]}"')
 
     exception = assert_raises do
       run_command("perform")
     end
-    assert_match "Health check against /up failed with status 500", exception.message
+    assert_match "container not ready (unhealthy)", exception.message
   end
 
   private

test/cli/main_test.rb

@@ -10,67 +10,140 @@ class CliMainTest < CliTestCase
   end
 
   test "deploy" do
-    invoke_options = { "config_file" => "test/fixtures/deploy_simple.yml", "skip_broadcast" => false, "version" => "999" }
+    invoke_options = { "config_file" => "test/fixtures/deploy_simple.yml", "version" => "999", "skip_hooks" => false }
 
-    Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:server:bootstrap", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:registry:login", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:build:deliver", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:traefik:boot", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:healthcheck:perform", [], invoke_options)
+    Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:app:stale_containers", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:app:boot", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:prune:all", [], invoke_options)
 
+    Mrsk::Commands::Hook.any_instance.stubs(:hook_exists?).returns(true)
+    hook_variables = { version: 999, service_version: "app@999", hosts: "1.1.1.1,1.1.1.2", command: "deploy" }
+
     run_command("deploy").tap do |output|
-      assert_match /Ensure curl and Docker are installed/, output
+      assert_hook_ran "pre-connect", output, **hook_variables
       assert_match /Log into image registry/, output
       assert_match /Build and push app image/, output
+      assert_hook_ran "pre-deploy", output, **hook_variables
       assert_match /Ensure Traefik is running/, output
       assert_match /Ensure app can pass healthcheck/, output
+      assert_match /Detect stale containers/, output
       assert_match /Prune old containers and images/, output
+      assert_hook_ran "post-deploy", output, **hook_variables, runtime: 0
     end
   end
 
   test "deploy with skip_push" do
-    invoke_options = { "config_file" => "test/fixtures/deploy_simple.yml", "skip_broadcast" => false, "version" => "999" }
+    invoke_options = { "config_file" => "test/fixtures/deploy_simple.yml", "version" => "999", "skip_hooks" => false }
 
-    Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:server:bootstrap", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:registry:login", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:build:pull", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:traefik:boot", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:healthcheck:perform", [], invoke_options)
+    Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:app:stale_containers", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:app:boot", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:prune:all", [], invoke_options)
 
     run_command("deploy", "--skip_push").tap do |output|
       assert_match /Acquiring the deploy lock/, output
-      assert_match /Ensure curl and Docker are installed/, output
       assert_match /Log into image registry/, output
       assert_match /Pull app image/, output
       assert_match /Ensure Traefik is running/, output
       assert_match /Ensure app can pass healthcheck/, output
+      assert_match /Detect stale containers/, output
       assert_match /Prune old containers and images/, output
       assert_match /Releasing the deploy lock/, output
     end
   end
 
+  test "deploy when locked" do
+    Thread.report_on_exception = false
+
+    SSHKit::Backend::Abstract.any_instance.stubs(:execute)
+      .with { |*arg| arg[0..1] == [:mkdir, :mrsk_lock] }
+      .raises(RuntimeError, "mkdir: cannot create directory ‘mrsk_lock’: File exists")
+
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_debug)
+      .with(:stat, :mrsk_lock, ">", "/dev/null", "&&", :cat, "mrsk_lock/details", "|", :base64, "-d")
+
+    assert_raises(Mrsk::Cli::LockError) do
+      run_command("deploy")
+    end
+  end
+
+  test "deploy error when locking" do
+    Thread.report_on_exception = false
+
+    SSHKit::Backend::Abstract.any_instance.stubs(:execute)
+      .with { |*arg| arg[0..1] == [:mkdir, :mrsk_lock] }
+      .raises(SocketError, "getaddrinfo: nodename nor servname provided, or not known")
+
+    assert_raises(SSHKit::Runner::ExecuteError) do
+      run_command("deploy")
+    end
+  end
+
+  test "deploy errors during outside section leave remove lock" do
+    invoke_options = { "config_file" => "test/fixtures/deploy_simple.yml", "version" => "999", "skip_hooks" => false }
+
+    Mrsk::Cli::Main.any_instance.expects(:invoke)
+      .with("mrsk:cli:registry:login", [], invoke_options)
+      .raises(RuntimeError)
+
+    assert !MRSK.holding_lock?
+    assert_raises(RuntimeError) do
+      stderred { run_command("deploy") }
+    end
+    assert !MRSK.holding_lock?
+  end
+
+  test "deploy with skipped hooks" do
+    invoke_options = { "config_file" => "test/fixtures/deploy_simple.yml", "version" => "999", "skip_hooks" => true }
+
+    Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:registry:login", [], invoke_options)
+    Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:build:deliver", [], invoke_options)
+    Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:traefik:boot", [], invoke_options)
+    Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:healthcheck:perform", [], invoke_options)
+    Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:app:stale_containers", [], invoke_options)
+    Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:app:boot", [], invoke_options)
+    Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:prune:all", [], invoke_options)
+
+    run_command("deploy", "--skip_hooks") do
+      refute_match /Running the post-deploy hook.../, output
+    end
+  end
+
   test "redeploy" do
-    invoke_options = { "config_file" => "test/fixtures/deploy_simple.yml", "skip_broadcast" => false, "version" => "999" }
+    invoke_options = { "config_file" => "test/fixtures/deploy_simple.yml", "version" => "999", "skip_hooks" => false }
 
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:build:deliver", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:healthcheck:perform", [], invoke_options)
+    Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:app:stale_containers", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:app:boot", [], invoke_options)
 
+    Mrsk::Commands::Hook.any_instance.stubs(:hook_exists?).returns(true)
+
+    hook_variables = { version: 999, service_version: "app@999", hosts: "1.1.1.1,1.1.1.2", command: "redeploy" }
+
     run_command("redeploy").tap do |output|
+      assert_hook_ran "pre-connect", output, **hook_variables
       assert_match /Build and push app image/, output
+      assert_hook_ran "pre-deploy", output, **hook_variables
+      assert_match /Running the pre-deploy hook.../, output
       assert_match /Ensure app can pass healthcheck/, output
+      assert_hook_ran "post-deploy", output, **hook_variables, runtime: "0"
     end
   end
 
   test "redeploy with skip_push" do
-    invoke_options = { "config_file" => "test/fixtures/deploy_simple.yml", "skip_broadcast" => false, "version" => "999" }
+    invoke_options = { "config_file" => "test/fixtures/deploy_simple.yml", "version" => "999", "skip_hooks" => false }
 
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:build:pull", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:healthcheck:perform", [], invoke_options)
+    Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:app:stale_containers", [], invoke_options)
     Mrsk::Cli::Main.any_instance.expects(:invoke).with("mrsk:cli:app:boot", [], invoke_options)
 
     run_command("redeploy", "--skip_push").tap do |output|
@@ -80,32 +153,63 @@ class CliMainTest < CliTestCase
   end
 
   test "rollback bad version" do
+    Thread.report_on_exception = false
+
     run_command("details") # Preheat MRSK const
 
     run_command("rollback", "nonsense").tap do |output|
-      assert_match /docker container ls --all --filter label=service=app --format '{{ .Names }}'/, output
+      assert_match /docker container ls --all --filter name=\^app-web-nonsense\$ --quiet/, output
       assert_match /The app version 'nonsense' is not available as a container/, output
     end
   end
 
   test "rollback good version" do
-    Mrsk::Cli::Main.any_instance.stubs(:container_name_available?).returns(true)
-    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info).with(:docker, :ps, "--filter", "label=service=app", "--format", "\"{{.Names}}\"", "|", "sed 's/-/\\n/g'", "|", "tail -n 1").returns("version-to-rollback\n").at_least_once
+    [ "web", "workers" ].each do |role|
+      SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+        .with(:docker, :container, :ls, "--filter", "name=^app-#{role}-123$", "--quiet", raise_on_non_zero_exit: false)
+        .returns("").at_least_once
+      SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+        .with(:docker, :container, :ls, "--all", "--filter", "name=^app-#{role}-123$", "--quiet")
+        .returns("version-to-rollback\n").at_least_once
+      SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+        .with(:docker, :ps, "--filter", "label=service=app", "--filter", "label=role=#{role}", "--filter", "status=running", "--filter", "status=restarting", "--latest", "--format", "\"{{.Names}}\"", "|", "grep -oE \"\\-[^-]+$\"", "|", "cut -c 2-", raise_on_non_zero_exit: false)
+        .returns("version-to-rollback\n").at_least_once
+      SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+        .with(:docker, :container, :ls, "--all", "--filter", "name=^app-#{role}-123$", "--quiet", "|", :xargs, :docker, :inspect, "--format", "'{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'")
+        .returns("running").at_least_once # health check
+    end
+
+    Mrsk::Commands::Hook.any_instance.stubs(:hook_exists?).returns(true)
+    hook_variables = { version: 123, service_version: "app@123", hosts: "1.1.1.1,1.1.1.2,1.1.1.3,1.1.1.4", command: "rollback" }
 
     run_command("rollback", "123", config_file: "deploy_with_accessories").tap do |output|
-      assert_match "Start version 123", output
-      assert_match "docker start app-123", output
-      assert_match "docker container ls --all --filter name=^app-version-to-rollback$ --quiet | xargs docker stop", output, "Should stop the container that was previously running"
+      assert_match "Start container with version 123", output
+      assert_hook_ran "pre-deploy", output, **hook_variables
+      assert_match "docker tag dhh/app:123 dhh/app:latest", output
+      assert_match "docker start app-web-123", output
+      assert_match "docker container ls --all --filter name=^app-web-version-to-rollback$ --quiet | xargs docker stop", output, "Should stop the container that was previously running"
+      assert_hook_ran "post-deploy", output, **hook_variables, runtime: "0"
     end
   end
 
   test "rollback without old version" do
-    Mrsk::Cli::Main.any_instance.stubs(:container_name_available?).returns(true)
-    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info).with(:docker, :ps, "--filter", "label=service=app", "--format", "\"{{.Names}}\"", "|", "sed 's/-/\\n/g'", "|", "tail -n 1").returns("").at_least_once
+    Mrsk::Cli::Main.any_instance.stubs(:container_available?).returns(true)
+
+    Mrsk::Utils::HealthcheckPoller.stubs(:sleep)
+
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:docker, :container, :ls, "--filter", "name=^app-web-123$", "--quiet", raise_on_non_zero_exit: false)
+      .returns("").at_least_once
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:docker, :ps, "--filter", "label=service=app", "--filter", "label=role=web", "--filter", "status=running", "--filter", "status=restarting", "--latest", "--format", "\"{{.Names}}\"", "|", "grep -oE \"\\-[^-]+$\"", "|", "cut -c 2-", raise_on_non_zero_exit: false)
+      .returns("").at_least_once
+    SSHKit::Backend::Abstract.any_instance.expects(:capture_with_info)
+      .with(:docker, :container, :ls, "--all", "--filter", "name=^app-web-123$", "--quiet", "|", :xargs, :docker, :inspect, "--format", "'{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'")
+      .returns("running").at_least_once # health check
 
     run_command("rollback", "123").tap do |output|
-      assert_match "Start version 123", output
-      assert_match "docker start app-123", output
+      assert_match "Start container with version 123", output
+      assert_match "docker start app-web-123 || docker run --detach --restart unless-stopped --name app-web-123", output
       assert_no_match "docker stop", output
     end
   end
@@ -165,9 +269,11 @@ class CliMainTest < CliTestCase
   end
 
   test "init" do
-    Pathname.any_instance.expects(:exist?).returns(false).twice
+    Pathname.any_instance.expects(:exist?).returns(false).times(3)
+    Pathname.any_instance.stubs(:mkpath)
     FileUtils.stubs(:mkdir_p)
     FileUtils.stubs(:cp_r)
+    FileUtils.stubs(:cp)
 
     run_command("init").tap do |output|
       assert_match /Created configuration file in config\/deploy.yml/, output
@@ -176,7 +282,7 @@ class CliMainTest < CliTestCase
   end
 
   test "init with existing config" do
-    Pathname.any_instance.expects(:exist?).returns(true).twice
+    Pathname.any_instance.expects(:exist?).returns(true).times(3)
 
     run_command("init").tap do |output|
       assert_match /Config file already exists in config\/deploy.yml \(remove first to create a new one\)/, output
@@ -184,9 +290,11 @@ class CliMainTest < CliTestCase
   end
 
   test "init with bundle option" do
-    Pathname.any_instance.expects(:exist?).returns(false).times(3)
+    Pathname.any_instance.expects(:exist?).returns(false).times(4)
+    Pathname.any_instance.stubs(:mkpath)
     FileUtils.stubs(:mkdir_p)
     FileUtils.stubs(:cp_r)
+    FileUtils.stubs(:cp)
 
     run_command("init", "--bundle").tap do |output|
       assert_match /Created configuration file in config\/deploy.yml/, output
@@ -199,9 +307,11 @@ class CliMainTest < CliTestCase
   end
 
   test "init with bundle option and existing binstub" do
-    Pathname.any_instance.expects(:exist?).returns(true).times(3)
+    Pathname.any_instance.expects(:exist?).returns(true).times(4)
+    Pathname.any_instance.stubs(:mkpath)
     FileUtils.stubs(:mkdir_p)
     FileUtils.stubs(:cp_r)
+    FileUtils.stubs(:cp)
 
     run_command("init", "--bundle").tap do |output|
       assert_match /Config file already exists in config\/deploy.yml \(remove first to create a new one\)/, output

test/cli/prune_test.rb

@@ -10,13 +10,14 @@ class CliPruneTest < CliTestCase
 
   test "images" do
     run_command("images").tap do |output|
-      assert_match /docker image prune --all --force --filter label=service=app --filter until=168h on 1.1.1.\d/, output
+      assert_match "docker image prune --force --filter label=service=app --filter dangling=true on 1.1.1.", output
+      assert_match "docker image ls --filter label=service=app --format '{{.ID}} {{.Repository}}:{{.Tag}}' | grep -v -w \"$(docker container ls -a --format '{{.Image}}\\|' --filter label=service=app | tr -d '\\n')dhh/app:latest\\|dhh/app:<none>\" | while read image tag; do docker rmi $tag; done on 1.1.1.", output
     end
   end
 
   test "containers" do
     run_command("containers").tap do |output|
-      assert_match /docker container prune --force --filter label=service=app --filter until=72h on 1.1.1.\d/, output
+      assert_match /docker ps -q -a --filter label=service=app --filter status=created --filter status=exited --filter status=dead | tail -n +6 | while read container_id; do docker rm $container_id; done on 1.1.1.\d/, output
     end
   end
 

test/cli/server_test.rb

@@ -1,11 +1,30 @@
 require_relative "cli_test_case"
 
 class CliServerTest < CliTestCase
-  test "bootstrap" do
+  test "bootstrap already installed" do
+    SSHKit::Backend::Abstract.any_instance.expects(:execute).with(:docker, "-v", raise_on_non_zero_exit: false).returns(true).at_least_once
+
+    assert_equal "", run_command("bootstrap")
+  end
+
+  test "bootstrap install as non-root user" do
+    SSHKit::Backend::Abstract.any_instance.expects(:execute).with(:docker, "-v", raise_on_non_zero_exit: false).returns(false).at_least_once
+    SSHKit::Backend::Abstract.any_instance.expects(:execute).with('[ "${EUID:-$(id -u)}" -eq 0 ]', raise_on_non_zero_exit: false).returns(false).at_least_once
+
+    assert_raise RuntimeError, "Docker is not installed on 1.1.1.1, 1.1.1.3, 1.1.1.4, 1.1.1.2 and can't be automatically installed without having root access and the `curl` command available. Install Docker manually: https://docs.docker.com/engine/install/" do
+      run_command("bootstrap")
+    end
+  end
+
+  test "bootstrap install as root user" do
+    SSHKit::Backend::Abstract.any_instance.expects(:execute).with(:docker, "-v", raise_on_non_zero_exit: false).returns(false).at_least_once
+    SSHKit::Backend::Abstract.any_instance.expects(:execute).with('[ "${EUID:-$(id -u)}" -eq 0 ]', raise_on_non_zero_exit: false).returns(true).at_least_once
+    SSHKit::Backend::Abstract.any_instance.expects(:execute).with(:curl, "-fsSL", "https://get.docker.com", "|", :sh).at_least_once
+
     run_command("bootstrap").tap do |output|
-      assert_match /which curl/, output
-      assert_match /which docker/, output
-      assert_match /apt-get update -y && apt-get install curl docker.io -y/, output
+      ("1.1.1.1".."1.1.1.4").map do |host|
+        assert_match "Missing Docker on #{host}. Installing…", output
+      end
     end
   end
 

test/cli/traefik_test.rb

@@ -3,7 +3,8 @@ require_relative "cli_test_case"
 class CliTraefikTest < CliTestCase
   test "boot" do
     run_command("boot").tap do |output|
-      assert_match "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" traefik:v2.9.9 --providers.docker --log.level=DEBUG", output
+      assert_match "docker login", output
+      assert_match "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" #{Mrsk::Commands::Traefik::DEFAULT_IMAGE} --providers.docker --log.level=DEBUG", output
     end
   end
 

test/commander_test.rb

@@ -2,9 +2,7 @@ require "test_helper"
 
 class CommanderTest < ActiveSupport::TestCase
   setup do
-    @mrsk = Mrsk::Commander.new.tap do |mrsk|
-      mrsk.configure config_file: Pathname.new(File.expand_path("fixtures/deploy_with_roles.yml", __dir__))
-    end
+    configure_with(:deploy_with_roles)
   end
 
   test "lazy configuration" do
@@ -47,12 +45,35 @@ class CommanderTest < ActiveSupport::TestCase
   end
 
   test "primary_host with specific hosts via role" do
-    @mrsk.specific_roles = "web"
-    assert_equal "1.1.1.1", @mrsk.primary_host
+    @mrsk.specific_roles = "workers"
+    assert_equal "1.1.1.3", @mrsk.primary_host
   end
 
   test "roles_on" do
     assert_equal [ "web" ], @mrsk.roles_on("1.1.1.1")
     assert_equal [ "workers" ], @mrsk.roles_on("1.1.1.3")
   end
+
+  test "default group strategy" do
+    assert_empty @mrsk.boot_strategy
+  end
+
+  test "specific limit group strategy" do
+    configure_with(:deploy_with_boot_strategy)
+
+    assert_equal({ in: :groups, limit: 3, wait: 2 }, @mrsk.boot_strategy)
+  end
+
+  test "percentage-based group strategy" do
+    configure_with(:deploy_with_percentage_boot_strategy)
+
+    assert_equal({ in: :groups, limit: 1, wait: 2 }, @mrsk.boot_strategy)
+  end
+
+  private
+    def configure_with(variant)
+      @mrsk = Mrsk::Commander.new.tap do |mrsk|
+        mrsk.configure config_file: Pathname.new(File.expand_path("fixtures/#{variant}.yml", __dir__))
+      end
+    end
 end

test/commands/app_test.rb

@@ -13,15 +13,21 @@ class CommandsAppTest < ActiveSupport::TestCase
 
   test "run" do
     assert_equal \
-      "docker run --detach --restart unless-stopped --name app-web-999 -e MRSK_CONTAINER_NAME=\"app-web-999\" -e RAILS_MASTER_KEY=\"456\" --log-opt max-size=\"10m\" --label service=\"app\" --label role=\"web\" --label traefik.http.routers.app.rule=\"PathPrefix(\\`/\\`)\" --label traefik.http.services.app.loadbalancer.healthcheck.path=\"/up\" --label traefik.http.services.app.loadbalancer.healthcheck.interval=\"1s\" --label traefik.http.middlewares.app-retry.retry.attempts=\"5\" --label traefik.http.middlewares.app-retry.retry.initialinterval=\"500ms\" --label traefik.http.routers.app.middlewares=\"app-retry@docker\" dhh/app:999",
+      "docker run --detach --restart unless-stopped --name app-web-999 -e MRSK_CONTAINER_NAME=\"app-web-999\" -e RAILS_MASTER_KEY=\"456\" --health-cmd \"curl -f http://localhost:3000/up || exit 1\" --health-interval \"1s\" --log-opt max-size=\"10m\" --label service=\"app\" --label role=\"web\" --label traefik.http.services.app-web.loadbalancer.server.scheme=\"http\" --label traefik.http.routers.app-web.rule=\"PathPrefix(\\`/\\`)\" --label traefik.http.middlewares.app-web-retry.retry.attempts=\"5\" --label traefik.http.middlewares.app-web-retry.retry.initialinterval=\"500ms\" --label traefik.http.routers.app-web.middlewares=\"app-web-retry@docker\" dhh/app:999",
       new_command.run.join(" ")
   end
 
+  test "run with hostname" do
+    assert_equal \
+      "docker run --detach --restart unless-stopped --name app-web-999 --hostname myhost -e MRSK_CONTAINER_NAME=\"app-web-999\" -e RAILS_MASTER_KEY=\"456\" --health-cmd \"curl -f http://localhost:3000/up || exit 1\" --health-interval \"1s\" --log-opt max-size=\"10m\" --label service=\"app\" --label role=\"web\" --label traefik.http.services.app-web.loadbalancer.server.scheme=\"http\" --label traefik.http.routers.app-web.rule=\"PathPrefix(\\`/\\`)\" --label traefik.http.middlewares.app-web-retry.retry.attempts=\"5\" --label traefik.http.middlewares.app-web-retry.retry.initialinterval=\"500ms\" --label traefik.http.routers.app-web.middlewares=\"app-web-retry@docker\" dhh/app:999",
+      new_command.run(hostname: "myhost").join(" ")
+  end
+
   test "run with volumes" do
     @config[:volumes] = ["/local/path:/container/path" ]
 
     assert_equal \
-      "docker run --detach --restart unless-stopped --name app-web-999 -e MRSK_CONTAINER_NAME=\"app-web-999\" -e RAILS_MASTER_KEY=\"456\" --log-opt max-size=\"10m\" --volume /local/path:/container/path --label service=\"app\" --label role=\"web\" --label traefik.http.routers.app.rule=\"PathPrefix(\\`/\\`)\" --label traefik.http.services.app.loadbalancer.healthcheck.path=\"/up\" --label traefik.http.services.app.loadbalancer.healthcheck.interval=\"1s\" --label traefik.http.middlewares.app-retry.retry.attempts=\"5\" --label traefik.http.middlewares.app-retry.retry.initialinterval=\"500ms\" --label traefik.http.routers.app.middlewares=\"app-retry@docker\" dhh/app:999",
+      "docker run --detach --restart unless-stopped --name app-web-999 -e MRSK_CONTAINER_NAME=\"app-web-999\" -e RAILS_MASTER_KEY=\"456\" --health-cmd \"curl -f http://localhost:3000/up || exit 1\" --health-interval \"1s\" --log-opt max-size=\"10m\" --volume /local/path:/container/path --label service=\"app\" --label role=\"web\" --label traefik.http.services.app-web.loadbalancer.server.scheme=\"http\" --label traefik.http.routers.app-web.rule=\"PathPrefix(\\`/\\`)\" --label traefik.http.middlewares.app-web-retry.retry.attempts=\"5\" --label traefik.http.middlewares.app-web-retry.retry.initialinterval=\"500ms\" --label traefik.http.routers.app-web.middlewares=\"app-web-retry@docker\" dhh/app:999",
       new_command.run.join(" ")
   end
 
@@ -29,7 +35,23 @@ class CommandsAppTest < ActiveSupport::TestCase
     @config[:healthcheck] = { "path" => "/healthz" }
 
     assert_equal \
-      "docker run --detach --restart unless-stopped --name app-web-999 -e MRSK_CONTAINER_NAME=\"app-web-999\" -e RAILS_MASTER_KEY=\"456\" --log-opt max-size=\"10m\" --label service=\"app\" --label role=\"web\" --label traefik.http.routers.app.rule=\"PathPrefix(\\`/\\`)\" --label traefik.http.services.app.loadbalancer.healthcheck.path=\"/healthz\" --label traefik.http.services.app.loadbalancer.healthcheck.interval=\"1s\" --label traefik.http.middlewares.app-retry.retry.attempts=\"5\" --label traefik.http.middlewares.app-retry.retry.initialinterval=\"500ms\" --label traefik.http.routers.app.middlewares=\"app-retry@docker\" dhh/app:999",
+      "docker run --detach --restart unless-stopped --name app-web-999 -e MRSK_CONTAINER_NAME=\"app-web-999\" -e RAILS_MASTER_KEY=\"456\" --health-cmd \"curl -f http://localhost:3000/healthz || exit 1\" --health-interval \"1s\" --log-opt max-size=\"10m\" --label service=\"app\" --label role=\"web\" --label traefik.http.services.app-web.loadbalancer.server.scheme=\"http\" --label traefik.http.routers.app-web.rule=\"PathPrefix(\\`/\\`)\" --label traefik.http.middlewares.app-web-retry.retry.attempts=\"5\" --label traefik.http.middlewares.app-web-retry.retry.initialinterval=\"500ms\" --label traefik.http.routers.app-web.middlewares=\"app-web-retry@docker\" dhh/app:999",
+      new_command.run.join(" ")
+  end
+
+  test "run with custom healthcheck command" do
+    @config[:healthcheck] = { "cmd" => "/bin/up" }
+
+    assert_equal \
+      "docker run --detach --restart unless-stopped --name app-web-999 -e MRSK_CONTAINER_NAME=\"app-web-999\" -e RAILS_MASTER_KEY=\"456\" --health-cmd \"/bin/up\" --health-interval \"1s\" --log-opt max-size=\"10m\" --label service=\"app\" --label role=\"web\" --label traefik.http.services.app-web.loadbalancer.server.scheme=\"http\" --label traefik.http.routers.app-web.rule=\"PathPrefix(\\`/\\`)\" --label traefik.http.middlewares.app-web-retry.retry.attempts=\"5\" --label traefik.http.middlewares.app-web-retry.retry.initialinterval=\"500ms\" --label traefik.http.routers.app-web.middlewares=\"app-web-retry@docker\" dhh/app:999",
+      new_command.run.join(" ")
+  end
+
+  test "run with role-specific healthcheck options" do
+    @config[:servers] = { "web" => { "hosts" => [ "1.1.1.1" ], "healthcheck" => { "cmd" => "/bin/healthy" } } }
+
+    assert_equal \
+      "docker run --detach --restart unless-stopped --name app-web-999 -e MRSK_CONTAINER_NAME=\"app-web-999\" -e RAILS_MASTER_KEY=\"456\" --health-cmd \"/bin/healthy\" --health-interval \"1s\" --log-opt max-size=\"10m\" --label service=\"app\" --label role=\"web\" --label traefik.http.services.app-web.loadbalancer.server.scheme=\"http\" --label traefik.http.routers.app-web.rule=\"PathPrefix(\\`/\\`)\" --label traefik.http.middlewares.app-web-retry.retry.attempts=\"5\" --label traefik.http.middlewares.app-web-retry.retry.initialinterval=\"500ms\" --label traefik.http.routers.app-web.middlewares=\"app-web-retry@docker\" dhh/app:999",
       new_command.run.join(" ")
   end
 
@@ -44,7 +66,7 @@ class CommandsAppTest < ActiveSupport::TestCase
     @config[:logging] = { "driver" => "local", "options" => { "max-size" => "100m", "max-file" => "3" } }
 
     assert_equal \
-      "docker run --detach --restart unless-stopped --name app-web-999 -e MRSK_CONTAINER_NAME=\"app-web-999\" -e RAILS_MASTER_KEY=\"456\" --log-driver \"local\" --log-opt max-size=\"100m\" --log-opt max-file=\"3\" --label service=\"app\" --label role=\"web\" --label traefik.http.routers.app.rule=\"PathPrefix(\\`/\\`)\" --label traefik.http.services.app.loadbalancer.healthcheck.path=\"/up\" --label traefik.http.services.app.loadbalancer.healthcheck.interval=\"1s\" --label traefik.http.middlewares.app-retry.retry.attempts=\"5\" --label traefik.http.middlewares.app-retry.retry.initialinterval=\"500ms\" --label traefik.http.routers.app.middlewares=\"app-retry@docker\" dhh/app:999",
+      "docker run --detach --restart unless-stopped --name app-web-999 -e MRSK_CONTAINER_NAME=\"app-web-999\" -e RAILS_MASTER_KEY=\"456\" --health-cmd \"curl -f http://localhost:3000/up || exit 1\" --health-interval \"1s\" --log-driver \"local\" --log-opt max-size=\"100m\" --log-opt max-file=\"3\" --label service=\"app\" --label role=\"web\" --label traefik.http.services.app-web.loadbalancer.server.scheme=\"http\" --label traefik.http.routers.app-web.rule=\"PathPrefix(\\`/\\`)\" --label traefik.http.middlewares.app-web-retry.retry.attempts=\"5\" --label traefik.http.middlewares.app-web-retry.retry.initialinterval=\"500ms\" --label traefik.http.routers.app-web.middlewares=\"app-web-retry@docker\" dhh/app:999",
       new_command.run.join(" ")
   end
 
@@ -61,16 +83,28 @@ class CommandsAppTest < ActiveSupport::TestCase
       new_command.start.join(" ")
   end
 
+  test "start_or_run" do
+    assert_equal \
+      "docker start app-web-999 || docker run --detach --restart unless-stopped --name app-web-999 -e MRSK_CONTAINER_NAME=\"app-web-999\" -e RAILS_MASTER_KEY=\"456\" --health-cmd \"curl -f http://localhost:3000/up || exit 1\" --health-interval \"1s\" --log-opt max-size=\"10m\" --label service=\"app\" --label role=\"web\" --label traefik.http.services.app-web.loadbalancer.server.scheme=\"http\" --label traefik.http.routers.app-web.rule=\"PathPrefix(\\`/\\`)\" --label traefik.http.middlewares.app-web-retry.retry.attempts=\"5\" --label traefik.http.middlewares.app-web-retry.retry.initialinterval=\"500ms\" --label traefik.http.routers.app-web.middlewares=\"app-web-retry@docker\" dhh/app:999",
+      new_command.start_or_run.join(" ")
+  end
+
+  test "start_or_run with hostname" do
+    assert_equal \
+      "docker start app-web-999 || docker run --detach --restart unless-stopped --name app-web-999 --hostname myhost -e MRSK_CONTAINER_NAME=\"app-web-999\" -e RAILS_MASTER_KEY=\"456\" --health-cmd \"curl -f http://localhost:3000/up || exit 1\" --health-interval \"1s\" --log-opt max-size=\"10m\" --label service=\"app\" --label role=\"web\" --label traefik.http.services.app-web.loadbalancer.server.scheme=\"http\" --label traefik.http.routers.app-web.rule=\"PathPrefix(\\`/\\`)\" --label traefik.http.middlewares.app-web-retry.retry.attempts=\"5\" --label traefik.http.middlewares.app-web-retry.retry.initialinterval=\"500ms\" --label traefik.http.routers.app-web.middlewares=\"app-web-retry@docker\" dhh/app:999",
+      new_command.start_or_run(hostname: "myhost").join(" ")
+  end
+
   test "stop" do
     assert_equal \
-      "docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker stop",
+      "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker stop",
       new_command.stop.join(" ")
   end
 
   test "stop with custom stop wait time" do
     @config[:stop_wait_time] = 30
     assert_equal \
-      "docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker stop -t 30",
+      "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker stop -t 30",
       new_command.stop.join(" ")
   end
 
@@ -96,37 +130,37 @@ class CommandsAppTest < ActiveSupport::TestCase
 
   test "logs" do
     assert_equal \
-      "docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker logs 2>&1",
+      "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs 2>&1",
       new_command.logs.join(" ")
 
     assert_equal \
-      "docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker logs --since 5m 2>&1",
+      "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs --since 5m 2>&1",
       new_command.logs(since: "5m").join(" ")
 
     assert_equal \
-      "docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker logs --tail 100 2>&1",
+      "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs --tail 100 2>&1",
       new_command.logs(lines: "100").join(" ")
 
     assert_equal \
-      "docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker logs --since 5m --tail 100 2>&1",
+      "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs --since 5m --tail 100 2>&1",
       new_command.logs(since: "5m", lines: "100").join(" ")
 
     assert_equal \
-      "docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker logs 2>&1 | grep 'my-id'",
+      "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs 2>&1 | grep 'my-id'",
       new_command.logs(grep: "my-id").join(" ")
 
     assert_equal \
-      "docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker logs --since 5m 2>&1 | grep 'my-id'",
+      "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs --since 5m 2>&1 | grep 'my-id'",
       new_command.logs(since: "5m", grep: "my-id").join(" ")
   end
 
   test "follow logs" do
     assert_match \
-      "docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker logs --timestamps --tail 10 --follow 2>&1",
+      "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs --timestamps --tail 10 --follow 2>&1",
       new_command.follow_logs(host: "app-1")
 
     assert_match \
-      "docker ps --quiet --filter label=service=app --filter label=role=web | xargs docker logs --timestamps --tail 10 --follow 2>&1 | grep \"Completed\"",
+      "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest | xargs docker logs --timestamps --tail 10 --follow 2>&1 | grep \"Completed\"",
       new_command.follow_logs(host: "app-1", grep: "Completed")
   end
 
@@ -178,17 +212,17 @@ class CommandsAppTest < ActiveSupport::TestCase
   end
 
 
-  test "current_container_id" do
+  test "current_running_container_id" do
     assert_equal \
-      "docker ps --quiet --filter label=service=app --filter label=role=web",
-      new_command.current_container_id.join(" ")
+      "docker ps --quiet --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest",
+      new_command.current_running_container_id.join(" ")
   end
 
-  test "current_container_id with destination" do
+  test "current_running_container_id with destination" do
     @destination = "staging"
     assert_equal \
-      "docker ps --quiet --filter label=service=app --filter label=destination=staging --filter label=role=web",
-      new_command.current_container_id.join(" ")
+      "docker ps --quiet --filter label=service=app --filter label=destination=staging --filter label=role=web --filter status=running --filter status=restarting --latest",
+      new_command.current_running_container_id.join(" ")
   end
 
   test "container_id_for" do
@@ -199,10 +233,20 @@ class CommandsAppTest < ActiveSupport::TestCase
 
   test "current_running_version" do
     assert_equal \
-      "docker ps --filter label=service=app --filter label=role=web --format \"{{.Names}}\" | sed 's/-/\\n/g' | tail -n 1",
+      "docker ps --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest --format \"{{.Names}}\" | grep -oE \"\\-[^-]+$\" | cut -c 2-",
       new_command.current_running_version.join(" ")
   end
 
+  test "list_versions" do
+    assert_equal \
+      "docker ps --filter label=service=app --filter label=role=web --format \"{{.Names}}\" | grep -oE \"\\-[^-]+$\" | cut -c 2-",
+      new_command.list_versions.join(" ")
+
+    assert_equal \
+      "docker ps --filter label=service=app --filter label=role=web --filter status=running --filter status=restarting --latest --format \"{{.Names}}\" | grep -oE \"\\-[^-]+$\" | cut -c 2-",
+      new_command.list_versions("--latest", statuses: [ :running, :restarting ]).join(" ")
+  end
+
   test "list_containers" do
     assert_equal \
       "docker container ls --all --filter label=service=app --filter label=role=web",
@@ -267,6 +311,12 @@ class CommandsAppTest < ActiveSupport::TestCase
       new_command.remove_images.join(" ")
   end
 
+  test "tag_current_as_latest" do
+    assert_equal \
+      "docker tag dhh/app:999 dhh/app:latest",
+      new_command.tag_current_as_latest.join(" ")
+  end
+
   private
     def new_command(role: "web")
       Mrsk::Commands::App.new(Mrsk::Configuration.new(@config, destination: @destination, version: "999"), role: role)

test/commands/auditor_test.rb

@@ -1,43 +1,64 @@
 require "test_helper"
+require "active_support/testing/time_helpers"
 
 class CommandsAuditorTest < ActiveSupport::TestCase
+  include ActiveSupport::Testing::TimeHelpers
+
   setup do
+    freeze_time
+
     @config = {
-      service: "app", image: "dhh/app", registry: { "username" => "dhh", "password" => "secret" }, servers: [ "1.1.1.1" ],
-      audit_broadcast_cmd: "bin/audit_broadcast"
+      service: "app", image: "dhh/app", registry: { "username" => "dhh", "password" => "secret" }, servers: [ "1.1.1.1" ]
     }
+
+    @auditor = new_command
+    @performer = `whoami`.strip
+    @recorded_at = Time.now.utc.iso8601
   end
 
   test "record" do
-    assert_match \
-      /echo '.* app removed container' >> mrsk-app-audit.log/,
-      new_command.record("app removed container").join(" ")
+    assert_equal [
+      :echo,
+      "[#{@recorded_at}] [#{@performer}]",
+      "app removed container",
+      ">>", "mrsk-app-audit.log"
+    ], @auditor.record("app removed container")
   end
 
   test "record with destination" do
-    @destination = "staging"
-
-    assert_match \
-      /echo '.* app removed container' >> mrsk-app-staging-audit.log/,
-      new_command.record("app removed container").join(" ")
+    new_command(destination: "staging").tap do |auditor|
+      assert_equal [
+        :echo,
+        "[#{@recorded_at}] [#{@performer}] [staging]",
+        "app removed container",
+        ">>", "mrsk-app-staging-audit.log"
+      ], auditor.record("app removed container")
+    end
   end
 
-  test "record with role" do
-    @role = "web"
-
-    assert_match \
-      /echo '.* \[web\] app removed container' >> mrsk-app-audit.log/,
-      new_command.record("app removed container").join(" ")
+  test "record with command details" do
+    new_command(role: "web").tap do |auditor|
+      assert_equal [
+        :echo,
+        "[#{@recorded_at}] [#{@performer}] [web]",
+        "app removed container",
+        ">>", "mrsk-app-audit.log"
+      ], auditor.record("app removed container")
+    end
   end
 
-  test "broadcast" do
-    assert_match \
-      /bin\/audit_broadcast '\[.*\] app removed container'/,
-      new_command.broadcast("app removed container").join(" ")
+  test "record with arg details" do
+    assert_equal [
+      :echo,
+      "[#{@recorded_at}] [#{@performer}] [value]",
+      "app removed container",
+      ">>", "mrsk-app-audit.log"
+    ], @auditor.record("app removed container", detail: "value")
   end
 
+
   private
-    def new_command
-      Mrsk::Commands::Auditor.new(Mrsk::Configuration.new(@config, destination: @destination, version: "123"), role: @role)
+    def new_command(destination: nil, **details)
+      Mrsk::Commands::Auditor.new(Mrsk::Configuration.new(@config, destination: destination, version: "123"), **details)
     end
 end

test/commands/builder_test.rb

@@ -52,12 +52,21 @@ class CommandsBuilderTest < ActiveSupport::TestCase
   end
 
   test "build dockerfile" do
+    Pathname.any_instance.expects(:exist?).returns(true).once
     builder = new_builder_command(builder: { "dockerfile" => "Dockerfile.xyz" })
     assert_equal \
       "-t dhh/app:123 -t dhh/app:latest --label service=\"app\" --file Dockerfile.xyz",
       builder.target.build_options.join(" ")
   end
 
+  test "missing dockerfile" do
+    Pathname.any_instance.expects(:exist?).returns(false).once
+    builder = new_builder_command(builder: { "dockerfile" => "Dockerfile.xyz" })
+    assert_raises(Mrsk::Commands::Builder::Base::BuilderError) do
+      builder.target.build_options.join(" ")
+    end
+  end
+
   test "build context" do
     builder = new_builder_command(builder: { "context" => ".." })
     assert_equal \
@@ -79,7 +88,7 @@ class CommandsBuilderTest < ActiveSupport::TestCase
       builder.push.join(" ")
   end
 
-  test "native push with with build secrets" do
+  test "native push with build secrets" do
     builder = new_builder_command(builder: { "multiarch" => false, "secrets" => [ "a", "b" ] })
     assert_equal \
       "docker build -t dhh/app:123 -t dhh/app:latest --label service=\"app\" --secret id=\"a\" --secret id=\"b\" --file Dockerfile . && docker push dhh/app:123 && docker push dhh/app:latest",

test/commands/docker_test.rb

@@ -0,0 +1,26 @@
+require "test_helper"
+
+class CommandsDockerTest < ActiveSupport::TestCase
+  setup do
+    @config = {
+      service: "app", image: "dhh/app", registry: { "username" => "dhh", "password" => "secret" }, servers: [ "1.1.1.1" ]
+    }
+    @docker = Mrsk::Commands::Docker.new(Mrsk::Configuration.new(@config))
+  end
+
+  test "install" do
+    assert_equal "curl -fsSL https://get.docker.com | sh", @docker.install.join(" ")
+  end
+
+  test "installed?" do
+    assert_equal "docker -v", @docker.installed?.join(" ")
+  end
+
+  test "running?" do
+    assert_equal "docker version", @docker.running?.join(" ")
+  end
+
+  test "superuser?" do
+    assert_equal '[ "${EUID:-$(id -u)}" -eq 0 ]', @docker.superuser?.join(" ")
+  end
+end

test/commands/healthcheck_test.rb

@@ -10,7 +10,7 @@ class CommandsHealthcheckTest < ActiveSupport::TestCase
 
   test "run" do
     assert_equal \
-      "docker run --detach --name healthcheck-app-123 --publish 3999:3000 --label service=healthcheck-app -e MRSK_CONTAINER_NAME=\"healthcheck-app\" dhh/app:123",
+      "docker run --detach --name healthcheck-app-123 --publish 3999:3000 --label service=healthcheck-app -e MRSK_CONTAINER_NAME=\"healthcheck-app\" --health-cmd \"curl -f http://localhost:3000/up || exit 1\" --health-interval \"1s\" dhh/app:123",
       new_command.run.join(" ")
   end
 
@@ -18,7 +18,7 @@ class CommandsHealthcheckTest < ActiveSupport::TestCase
     @config[:healthcheck] = { "port" => 3001 }
 
     assert_equal \
-      "docker run --detach --name healthcheck-app-123 --publish 3999:3001 --label service=healthcheck-app -e MRSK_CONTAINER_NAME=\"healthcheck-app\" dhh/app:123",
+      "docker run --detach --name healthcheck-app-123 --publish 3999:3001 --label service=healthcheck-app -e MRSK_CONTAINER_NAME=\"healthcheck-app\" --health-cmd \"curl -f http://localhost:3001/up || exit 1\" --health-interval \"1s\" dhh/app:123",
       new_command.run.join(" ")
   end
 
@@ -26,29 +26,35 @@ class CommandsHealthcheckTest < ActiveSupport::TestCase
     @destination = "staging"
 
     assert_equal \
-      "docker run --detach --name healthcheck-app-staging-123 --publish 3999:3000 --label service=healthcheck-app-staging -e MRSK_CONTAINER_NAME=\"healthcheck-app-staging\" dhh/app:123",
+      "docker run --detach --name healthcheck-app-staging-123 --publish 3999:3000 --label service=healthcheck-app-staging -e MRSK_CONTAINER_NAME=\"healthcheck-app-staging\" --health-cmd \"curl -f http://localhost:3000/up || exit 1\" --health-interval \"1s\" dhh/app:123",
+      new_command.run.join(" ")
+  end
+
+  test "run with custom healthcheck" do
+    @config[:healthcheck] = { "cmd" => "/bin/up" }
+
+    assert_equal \
+      "docker run --detach --name healthcheck-app-123 --publish 3999:3000 --label service=healthcheck-app -e MRSK_CONTAINER_NAME=\"healthcheck-app\" --health-cmd \"/bin/up\" --health-interval \"1s\" dhh/app:123",
       new_command.run.join(" ")
   end
 
   test "run with custom options" do
     @config[:servers] = { "web" => { "hosts" => [ "1.1.1.1" ], "options" => { "mount" => "somewhere" } } }
     assert_equal \
-      "docker run --detach --name healthcheck-app-123 --publish 3999:3000 --label service=healthcheck-app -e MRSK_CONTAINER_NAME=\"healthcheck-app\" --mount \"somewhere\" dhh/app:123",
+      "docker run --detach --name healthcheck-app-123 --publish 3999:3000 --label service=healthcheck-app -e MRSK_CONTAINER_NAME=\"healthcheck-app\" --health-cmd \"curl -f http://localhost:3000/up || exit 1\" --health-interval \"1s\" --mount \"somewhere\" dhh/app:123",
       new_command.run.join(" ")
   end
 
-  test "curl" do
+  test "status" do
     assert_equal \
-      "curl --silent --output /dev/null --write-out '%{http_code}' --max-time 2 http://localhost:3999/up",
-      new_command.curl.join(" ")
+      "docker container ls --all --filter name=^healthcheck-app-123$ --quiet | xargs docker inspect --format '{{if .State.Health}}{{.State.Health.Status}}{{else}}{{.State.Status}}{{end}}'",
+      new_command.status.join(" ")
   end
 
-  test "curl with custom path" do
-    @config[:healthcheck] = { "path" => "/healthz" }
-
+  test "container_health_log" do
     assert_equal \
-      "curl --silent --output /dev/null --write-out '%{http_code}' --max-time 2 http://localhost:3999/healthz",
-      new_command.curl.join(" ")
+      "docker container ls --all --filter name=^healthcheck-app-123$ --quiet | xargs docker inspect --format '{{json .State.Health}}'",
+      new_command.container_health_log.join(" ")
   end
 
   test "stop" do

test/commands/hook_test.rb

@@ -0,0 +1,44 @@
+require "test_helper"
+
+class CommandsHookTest < ActiveSupport::TestCase
+  include ActiveSupport::Testing::TimeHelpers
+
+  setup do
+    freeze_time
+
+    @config = {
+      service: "app", image: "dhh/app", registry: { "username" => "dhh", "password" => "secret" }, servers: [ "1.1.1.1" ],
+      traefik: { "args" => { "accesslog.format" => "json", "metrics.prometheus.buckets" => "0.1,0.3,1.2,5.0" } }
+    }
+
+    @performer = `whoami`.strip
+    @recorded_at = Time.now.utc.iso8601
+  end
+
+  test "run" do
+    assert_equal [
+      ".mrsk/hooks/foo",
+      { env: {
+        "MRSK_RECORDED_AT" => @recorded_at,
+        "MRSK_PERFORMER" => @performer,
+        "MRSK_VERSION" => "123",
+        "MRSK_SERVICE_VERSION" => "app@123" } }
+    ], new_command.run("foo")
+  end
+
+  test "run with custom hooks_path" do
+    assert_equal [
+      "custom/hooks/path/foo",
+      { env: {
+        "MRSK_RECORDED_AT" => @recorded_at,
+        "MRSK_PERFORMER" => @performer,
+        "MRSK_VERSION" => "123",
+        "MRSK_SERVICE_VERSION" => "app@123" } }
+    ], new_command(hooks_path: "custom/hooks/path").run("foo")
+  end
+
+  private
+    def new_command(**extra_config)
+      Mrsk::Commands::Hook.new(Mrsk::Configuration.new(@config.merge(**extra_config), version: "123"))
+    end
+end

test/commands/prune_test.rb

@@ -8,15 +8,21 @@ class CommandsPruneTest < ActiveSupport::TestCase
     }
   end
 
-  test "images" do
+  test "dangling images" do
     assert_equal \
-      "docker image prune --all --force --filter label=service=app --filter until=168h",
-      new_command.images.join(" ")
+      "docker image prune --force --filter label=service=app --filter dangling=true",
+      new_command.dangling_images.join(" ")
+  end
+
+  test "tagged images" do
+    assert_equal \
+      "docker image ls --filter label=service=app --format '{{.ID}} {{.Repository}}:{{.Tag}}' | grep -v -w \"$(docker container ls -a --format '{{.Image}}\\|' --filter label=service=app | tr -d '\\n')dhh/app:latest\\|dhh/app:<none>\" | while read image tag; do docker rmi $tag; done",
+      new_command.tagged_images.join(" ")
   end
 
   test "containers" do
     assert_equal \
-      "docker container prune --force --filter label=service=app --filter until=72h",
+      "docker ps -q -a --filter label=service=app --filter status=created --filter status=exited --filter status=dead | tail -n +6 | while read container_id; do docker rm $container_id; done",
       new_command.containers.join(" ")
   end
 

test/commands/traefik_test.rb

@@ -2,53 +2,83 @@ require "test_helper"
 
 class CommandsTraefikTest < ActiveSupport::TestCase
   setup do
+    @image = "traefik:test"
+
     @config = {
       service: "app", image: "dhh/app", registry: { "username" => "dhh", "password" => "secret" }, servers: [ "1.1.1.1" ],
-      traefik: { "args" => { "accesslog.format" => "json", "api.insecure" => true, "metrics.prometheus.buckets" => "0.1,0.3,1.2,5.0" } }
+      traefik: { "image" => @image, "args" => { "accesslog.format" => "json", "api.insecure" => true, "metrics.prometheus.buckets" => "0.1,0.3,1.2,5.0" } }
     }
+
+    ENV["EXAMPLE_API_KEY"] = "456"
+  end
+
+  teardown do
+    ENV.delete("EXAMPLE_API_KEY")
   end
 
   test "run" do
     assert_equal \
-      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" traefik:v2.9.9 --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
+      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" #{@image} --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
       new_command.run.join(" ")
 
     @config[:traefik]["host_port"] = "8080"
     assert_equal \
-      "docker run --name traefik --detach --restart unless-stopped --publish 8080:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" traefik:v2.9.9 --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
+      "docker run --name traefik --detach --restart unless-stopped --publish 8080:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" #{@image} --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
       new_command.run.join(" ")
   end
 
   test "run with ports configured" do
     assert_equal \
-      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" traefik:v2.9.9 --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
+      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" #{@image} --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
       new_command.run.join(" ")
 
     @config[:traefik]["options"] = {"publish" => %w[9000:9000 9001:9001]}
     assert_equal \
-      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" --publish \"9000:9000\" --publish \"9001:9001\" traefik:v2.9.9 --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
+      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" --publish \"9000:9000\" --publish \"9001:9001\" #{@image} --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
       new_command.run.join(" ")
   end
 
   test "run with volumes configured" do
     assert_equal \
-      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" traefik:v2.9.9 --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
+      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" #{@image} --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
       new_command.run.join(" ")
 
     @config[:traefik]["options"] = {"volume" => %w[./letsencrypt/acme.json:/letsencrypt/acme.json] }
     assert_equal \
-      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" --volume \"./letsencrypt/acme.json:/letsencrypt/acme.json\" traefik:v2.9.9 --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
+      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" --volume \"./letsencrypt/acme.json:/letsencrypt/acme.json\" #{@image} --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
       new_command.run.join(" ")
   end
 
   test "run with several options configured" do
     assert_equal \
-      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" traefik:v2.9.9 --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
+      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" #{@image} --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
       new_command.run.join(" ")
 
     @config[:traefik]["options"] = {"volume" => %w[./letsencrypt/acme.json:/letsencrypt/acme.json], "publish" => %w[8080:8080], "memory" => "512m"}
     assert_equal \
-      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" --volume \"./letsencrypt/acme.json:/letsencrypt/acme.json\" --publish \"8080:8080\" --memory \"512m\" traefik:v2.9.9 --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
+      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" --volume \"./letsencrypt/acme.json:/letsencrypt/acme.json\" --publish \"8080:8080\" --memory \"512m\" #{@image} --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
+      new_command.run.join(" ")
+  end
+
+  test "run with labels configured" do
+    assert_equal \
+      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" #{@image} --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
+      new_command.run.join(" ")
+
+    @config[:traefik]["labels"] = { "traefik.http.routers.dashboard.service" => "api@internal", "traefik.http.routers.dashboard.middlewares" => "auth" }
+    assert_equal \
+      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" --label traefik.http.routers.dashboard.service=\"api@internal\" --label traefik.http.routers.dashboard.middlewares=\"auth\" #{@image} --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
+      new_command.run.join(" ")
+  end
+
+  test "run with env configured" do
+    assert_equal \
+      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" #{@image} --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
+      new_command.run.join(" ")
+
+    @config[:traefik]["env"] = { "secret" => %w[EXAMPLE_API_KEY] }
+    assert_equal \
+      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock -e EXAMPLE_API_KEY=\"456\" --log-opt max-size=\"10m\" #{@image} --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
       new_command.run.join(" ")
   end
 
@@ -56,7 +86,7 @@ class CommandsTraefikTest < ActiveSupport::TestCase
     @config.delete(:traefik)
 
     assert_equal \
-      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" traefik:v2.9.9 --providers.docker --log.level=DEBUG",
+      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-opt max-size=\"10m\" #{Mrsk::Commands::Traefik::DEFAULT_IMAGE} --providers.docker --log.level=DEBUG",
       new_command.run.join(" ")
   end
 
@@ -64,7 +94,7 @@ class CommandsTraefikTest < ActiveSupport::TestCase
     @config[:logging] = { "driver" => "local", "options" => { "max-size" => "100m", "max-file" => "3" } }
 
     assert_equal \
-      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-driver \"local\" --log-opt max-size=\"100m\" --log-opt max-file=\"3\" traefik:v2.9.9 --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
+      "docker run --name traefik --detach --restart unless-stopped --publish 80:80 --volume /var/run/docker.sock:/var/run/docker.sock --log-driver \"local\" --log-opt max-size=\"100m\" --log-opt max-file=\"3\" #{@image} --providers.docker --log.level=DEBUG --accesslog.format=\"json\" --api.insecure --metrics.prometheus.buckets=\"0.1,0.3,1.2,5.0\"",
       new_command.run.join(" ")
   end
 

test/configuration/accessory_test.rb

@@ -112,8 +112,11 @@ class ConfigurationAccessoryTest < ActiveSupport::TestCase
 
   test "env args with secret" do
     ENV["MYSQL_ROOT_PASSWORD"] = "secret123"
-    assert_equal ["-e", "MYSQL_ROOT_PASSWORD=\"secret123\"", "-e", "MYSQL_ROOT_HOST=\"%\""], @config.accessory(:mysql).env_args
-    assert @config.accessory(:mysql).env_args[1].is_a?(SSHKit::Redaction)
+
+    @config.accessory(:mysql).env_args.tap do |env_args|
+      assert_equal ["-e", "MYSQL_ROOT_PASSWORD=\"secret123\"", "-e", "MYSQL_ROOT_HOST=\"%\""], Mrsk::Utils.unredacted(env_args)
+      assert_equal ["-e", "MYSQL_ROOT_PASSWORD=[REDACTED]", "-e", "MYSQL_ROOT_HOST=\"%\""], Mrsk::Utils.redacted(env_args)
+    end
   ensure
     ENV["MYSQL_ROOT_PASSWORD"] = nil
   end

test/configuration/role_test.rb

@@ -42,7 +42,7 @@ class ConfigurationRoleTest < ActiveSupport::TestCase
   end
 
   test "special label args for web" do
-    assert_equal [ "--label", "service=\"app\"", "--label", "role=\"web\"", "--label", "traefik.http.routers.app.rule=\"PathPrefix(\\`/\\`)\"", "--label", "traefik.http.services.app.loadbalancer.healthcheck.path=\"/up\"", "--label", "traefik.http.services.app.loadbalancer.healthcheck.interval=\"1s\"", "--label", "traefik.http.middlewares.app-retry.retry.attempts=\"5\"", "--label", "traefik.http.middlewares.app-retry.retry.initialinterval=\"500ms\"", "--label", "traefik.http.routers.app.middlewares=\"app-retry@docker\"" ], @config.role(:web).label_args
+    assert_equal [ "--label", "service=\"app\"", "--label", "role=\"web\"", "--label", "traefik.http.services.app-web.loadbalancer.server.scheme=\"http\"", "--label", "traefik.http.routers.app-web.rule=\"PathPrefix(\\`/\\`)\"", "--label", "traefik.http.middlewares.app-web-retry.retry.attempts=\"5\"", "--label", "traefik.http.middlewares.app-web-retry.retry.initialinterval=\"500ms\"", "--label", "traefik.http.routers.app-web.middlewares=\"app-web-retry@docker\"" ], @config.role(:web).label_args
   end
 
   test "custom labels" do
@@ -57,8 +57,8 @@ class ConfigurationRoleTest < ActiveSupport::TestCase
   end
 
   test "overwriting default traefik label" do
-    @deploy[:labels] = { "traefik.http.routers.app.rule" => "\"Host(\\`example.com\\`) || (Host(\\`example.org\\`) && Path(\\`/traefik\\`))\"" }
-    assert_equal "\"Host(\\`example.com\\`) || (Host(\\`example.org\\`) && Path(\\`/traefik\\`))\"", @config.role(:web).labels["traefik.http.routers.app.rule"]
+    @deploy[:labels] = { "traefik.http.routers.app-web.rule" => "\"Host(\\`example.com\\`) || (Host(\\`example.org\\`) && Path(\\`/traefik\\`))\"" }
+    assert_equal "\"Host(\\`example.com\\`) || (Host(\\`example.org\\`) && Path(\\`/traefik\\`))\"", @config.role(:web).labels["traefik.http.routers.app-web.rule"]
   end
 
   test "default traefik label on non-web role" do
@@ -66,7 +66,7 @@ class ConfigurationRoleTest < ActiveSupport::TestCase
       c[:servers]["beta"] = { "traefik" => "true", "hosts" => [ "1.1.1.5" ] }
     })
 
-    assert_equal [ "--label", "service=\"app\"", "--label", "role=\"beta\"", "--label", "traefik.http.routers.app.rule=\"PathPrefix(\\`/\\`)\"", "--label", "traefik.http.services.app.loadbalancer.healthcheck.path=\"/up\"", "--label", "traefik.http.services.app.loadbalancer.healthcheck.interval=\"1s\"", "--label", "traefik.http.middlewares.app-retry.retry.attempts=\"5\"", "--label", "traefik.http.middlewares.app-retry.retry.initialinterval=\"500ms\"", "--label", "traefik.http.routers.app.middlewares=\"app-retry@docker\"" ], config.role(:beta).label_args
+    assert_equal [ "--label", "service=\"app\"", "--label", "role=\"beta\"", "--label", "traefik.http.services.app-beta.loadbalancer.server.scheme=\"http\"", "--label", "traefik.http.routers.app-beta.rule=\"PathPrefix(\\`/\\`)\"", "--label", "traefik.http.middlewares.app-beta-retry.retry.attempts=\"5\"", "--label", "traefik.http.middlewares.app-beta-retry.retry.initialinterval=\"500ms\"", "--label", "traefik.http.routers.app-beta.middlewares=\"app-beta-retry@docker\"" ], config.role(:beta).label_args
   end
 
   test "env overwritten by role" do
@@ -97,7 +97,10 @@ class ConfigurationRoleTest < ActiveSupport::TestCase
     ENV["REDIS_PASSWORD"] = "secret456"
     ENV["DB_PASSWORD"] = "secret&\"123"
 
-    assert_equal ["-e", "REDIS_PASSWORD=\"secret456\"", "-e", "DB_PASSWORD=\"secret&\\\"123\"", "-e", "REDIS_URL=\"redis://a/b\"", "-e", "WEB_CONCURRENCY=\"4\""], @config_with_roles.role(:workers).env_args
+    @config_with_roles.role(:workers).env_args.tap do |env_args|
+      assert_equal ["-e", "REDIS_PASSWORD=\"secret456\"", "-e", "DB_PASSWORD=\"secret&\\\"123\"", "-e", "REDIS_URL=\"redis://a/b\"", "-e", "WEB_CONCURRENCY=\"4\""], Mrsk::Utils.unredacted(env_args)
+      assert_equal ["-e", "REDIS_PASSWORD=[REDACTED]", "-e", "DB_PASSWORD=[REDACTED]", "-e", "REDIS_URL=\"redis://a/b\"", "-e", "WEB_CONCURRENCY=\"4\""], Mrsk::Utils.redacted(env_args)
+    end
   ensure
     ENV["REDIS_PASSWORD"] = nil
     ENV["DB_PASSWORD"] = nil
@@ -116,7 +119,10 @@ class ConfigurationRoleTest < ActiveSupport::TestCase
 
     ENV["DB_PASSWORD"] = "secret123"
 
-    assert_equal ["-e", "DB_PASSWORD=\"secret123\"", "-e", "REDIS_URL=\"redis://a/b\"", "-e", "WEB_CONCURRENCY=\"4\""], @config_with_roles.role(:workers).env_args
+    @config_with_roles.role(:workers).env_args.tap do |env_args|
+      assert_equal ["-e", "DB_PASSWORD=\"secret123\"", "-e", "REDIS_URL=\"redis://a/b\"", "-e", "WEB_CONCURRENCY=\"4\""], Mrsk::Utils.unredacted(env_args)
+      assert_equal ["-e", "DB_PASSWORD=[REDACTED]", "-e", "REDIS_URL=\"redis://a/b\"", "-e", "WEB_CONCURRENCY=\"4\""], Mrsk::Utils.redacted(env_args)
+    end
   ensure
     ENV["DB_PASSWORD"] = nil
   end
@@ -133,7 +139,10 @@ class ConfigurationRoleTest < ActiveSupport::TestCase
 
     ENV["REDIS_PASSWORD"] = "secret456"
 
-    assert_equal ["-e", "REDIS_PASSWORD=\"secret456\"", "-e", "REDIS_URL=\"redis://a/b\"", "-e", "WEB_CONCURRENCY=\"4\""], @config_with_roles.role(:workers).env_args
+    @config_with_roles.role(:workers).env_args.tap do |env_args|
+      assert_equal ["-e", "REDIS_PASSWORD=\"secret456\"", "-e", "REDIS_URL=\"redis://a/b\"", "-e", "WEB_CONCURRENCY=\"4\""], Mrsk::Utils.unredacted(env_args)
+      assert_equal ["-e", "REDIS_PASSWORD=[REDACTED]", "-e", "REDIS_URL=\"redis://a/b\"", "-e", "WEB_CONCURRENCY=\"4\""], Mrsk::Utils.redacted(env_args)
+    end
   ensure
     ENV["REDIS_PASSWORD"] = nil
   end

test/configuration_test.rb

@@ -48,7 +48,7 @@ class ConfigurationTest < ActiveSupport::TestCase
   end
 
   test "role" do
-    assert_equal "web", @config.role(:web).name
+    assert @config.role(:web).name.web?
     assert_equal "workers", @config_with_roles.role(:workers).name
     assert_nil @config.role(:missing)
   end
@@ -72,19 +72,36 @@ class ConfigurationTest < ActiveSupport::TestCase
     assert_equal [ "1.1.1.1", "1.1.1.2", "1.1.1.3" ], config.traefik_hosts
   end
 
-  test "version" do
+  test "version no git repo" do
     ENV.delete("VERSION")
 
     @config.expects(:system).with("git rev-parse").returns(nil)
     error = assert_raises(RuntimeError) { @config.version}
     assert_match /no git repository found/, error.message
+  end
 
-    @config.expects(:current_commit_hash).returns("git-version")
+  test "version from git committed" do
+    ENV.delete("VERSION")
+
+    @config.expects(:`).with("git rev-parse HEAD").returns("git-version")
+    @config.expects(:`).with("git status --porcelain").returns("")
     assert_equal "git-version", @config.version
+  end
 
+  test "version from git uncommitted" do
+    ENV.delete("VERSION")
+
+    @config.expects(:`).with("git rev-parse HEAD").returns("git-version")
+    @config.expects(:`).with("git status --porcelain").returns("M   file\n")
+    assert_match /^git-version_uncommitted_[0-9a-f]{16}$/, @config.version
+  end
+
+  test "version from env" do
     ENV["VERSION"] = "env-version"
     assert_equal "env-version", @config.version
+  end
 
+  test "version from arg" do
     @config.version = "arg-version"
     assert_equal "arg-version", @config.version
   end
@@ -113,12 +130,13 @@ class ConfigurationTest < ActiveSupport::TestCase
 
   test "env args with clear and secrets" do
     ENV["PASSWORD"] = "secret123"
+
     config = Mrsk::Configuration.new(@deploy.tap { |c| c.merge!({
       env: { "clear" => { "PORT" => "3000" }, "secret" => [ "PASSWORD" ] }
     }) })
 
-    assert_equal [ "-e", "PASSWORD=\"secret123\"", "-e", "PORT=\"3000\"" ], config.env_args
-    assert config.env_args[1].is_a?(SSHKit::Redaction)
+    assert_equal [ "-e", "PASSWORD=\"secret123\"", "-e", "PORT=\"3000\"" ], Mrsk::Utils.unredacted(config.env_args)
+    assert_equal [ "-e", "PASSWORD=[REDACTED]", "-e", "PORT=\"3000\"" ], Mrsk::Utils.redacted(config.env_args)
   ensure
     ENV["PASSWORD"] = nil
   end
@@ -133,12 +151,13 @@ class ConfigurationTest < ActiveSupport::TestCase
 
   test "env args with only secrets" do
     ENV["PASSWORD"] = "secret123"
+
     config = Mrsk::Configuration.new(@deploy.tap { |c| c.merge!({
       env: { "secret" => [ "PASSWORD" ] }
     }) })
 
-    assert_equal [ "-e", "PASSWORD=\"secret123\"" ], config.env_args
-    assert config.env_args[1].is_a?(SSHKit::Redaction)
+    assert_equal [ "-e", "PASSWORD=\"secret123\"" ], Mrsk::Utils.unredacted(config.env_args)
+    assert_equal [ "-e", "PASSWORD=[REDACTED]" ], Mrsk::Utils.redacted(config.env_args)
   ensure
     ENV["PASSWORD"] = nil
   end
@@ -247,6 +266,6 @@ class ConfigurationTest < ActiveSupport::TestCase
   end
 
   test "to_h" do
-    assert_equal({ :roles=>["web"], :hosts=>["1.1.1.1", "1.1.1.2"], :primary_host=>"1.1.1.1", :version=>"missing", :repository=>"dhh/app", :absolute_image=>"dhh/app:missing", :service_with_version=>"app-missing", :env_args=>["-e", "REDIS_URL=\"redis://x/y\""], :ssh_options=>{:user=>"root", :auth_methods=>["publickey"]}, :volume_args=>["--volume", "/local/path:/container/path"], :logging=>["--log-opt", "max-size=\"10m\""], :healthcheck=>{"path"=>"/up", "port"=>3000 }}, @config.to_h)
+    assert_equal({ :roles=>["web"], :hosts=>["1.1.1.1", "1.1.1.2"], :primary_host=>"1.1.1.1", :version=>"missing", :repository=>"dhh/app", :absolute_image=>"dhh/app:missing", :service_with_version=>"app-missing", :env_args=>["-e", "REDIS_URL=\"redis://x/y\""], :ssh_options=>{:user=>"root", :auth_methods=>["publickey"]}, :volume_args=>["--volume", "/local/path:/container/path"], :logging=>["--log-opt", "max-size=\"10m\""], :healthcheck=>{"path"=>"/up", "port"=>3000, "max_attempts" => 7 }}, @config.to_h)
   end
 end

test/fixtures/deploy_with_boot_strategy.yml

@@ -0,0 +1,17 @@
+service: app
+image: dhh/app
+servers:
+  web:
+    - "1.1.1.1"
+    - "1.1.1.2"
+  workers:
+    - "1.1.1.3"
+    - "1.1.1.4"
+
+registry:
+  username: user
+  password: pw
+
+boot:
+  limit: 3
+  wait: 2

test/fixtures/deploy_with_percentage_boot_strategy.yml

@@ -0,0 +1,17 @@
+service: app
+image: dhh/app
+servers:
+  web:
+    - "1.1.1.1"
+    - "1.1.1.2"
+  workers:
+    - "1.1.1.3"
+    - "1.1.1.4"
+
+registry:
+  username: user
+  password: pw
+
+boot:
+  limit: 25%
+  wait: 2

test/integration/accessory_test.rb

@@ -0,0 +1,36 @@
+require_relative "integration_test"
+
+class AccessoryTest < IntegrationTest
+  test "boot, stop, start, restart, logs, remove" do
+    mrsk :accessory, :boot, :busybox
+    assert_accessory_running :busybox
+
+    mrsk :accessory, :stop, :busybox
+    assert_accessory_not_running :busybox
+
+    mrsk :accessory, :start, :busybox
+    assert_accessory_running :busybox
+
+    mrsk :accessory, :restart, :busybox
+    assert_accessory_running :busybox
+
+    logs = mrsk :accessory, :logs, :busybox, capture: true
+    assert_match /Starting busybox.../, logs
+
+    mrsk :accessory, :remove, :busybox, "-y"
+    assert_accessory_not_running :busybox
+  end
+
+  private
+    def assert_accessory_running(name)
+      assert_match /registry:4443\/busybox:1.36.0   "sh -c 'echo \\"Start/, accessory_details(name)
+    end
+
+    def assert_accessory_not_running(name)
+      refute_match /registry:4443\/busybox:1.36.0   "sh -c 'echo \\"Start/, accessory_details(name)
+    end
+
+    def accessory_details(name)
+      mrsk :accessory, :details, name, capture: true
+    end
+end

test/integration/app_test.rb

@@ -0,0 +1,55 @@
+require_relative "integration_test"
+
+class AppTest < IntegrationTest
+  test "stop, start, boot, logs, images, containers, exec, remove" do
+    mrsk :deploy
+
+    assert_app_is_up
+
+    mrsk :app, :stop
+
+    # traefik is up and returns 404s when it can't match a route
+    assert_app_not_found
+
+    mrsk :app, :start
+
+    # mrsk app start does not wait
+    wait_for_app_to_be_up
+
+    mrsk :app, :boot
+
+    wait_for_app_to_be_up
+
+    logs = mrsk :app, :logs, capture: true
+    assert_match /App Host: vm1/, logs
+    assert_match /App Host: vm2/, logs
+    assert_match /GET \/ HTTP\/1.1/, logs
+
+    images = mrsk :app, :images, capture: true
+    assert_match /App Host: vm1/, images
+    assert_match /App Host: vm2/, images
+    assert_match /registry:4443\/app\s+#{latest_app_version}/, images
+    assert_match /registry:4443\/app\s+latest/, images
+
+    containers = mrsk :app, :containers, capture: true
+    assert_match /App Host: vm1/, containers
+    assert_match /App Host: vm2/, containers
+    assert_match /registry:4443\/app:#{latest_app_version}/, containers
+    assert_match /registry:4443\/app:latest/, containers
+
+    exec_output = mrsk :app, :exec, :ps, capture: true
+    assert_match /App Host: vm1/, exec_output
+    assert_match /App Host: vm2/, exec_output
+    assert_match /1 root      0:\d\d ps/, exec_output
+
+    exec_output = mrsk :app, :exec, "--reuse", :ps, capture: true
+    assert_match /App Host: vm1/, exec_output
+    assert_match /App Host: vm2/, exec_output
+    assert_match /1 root      0:\d\d nginx/, exec_output
+
+    mrsk :app, :remove
+
+    # traefik is up and returns 404s when it can't match a route
+    assert_app_not_found
+  end
+end

test/integration/docker-compose.yml

@@ -0,0 +1,57 @@
+version: "3.7"
+name: "mrsk-test"
+
+volumes:
+  shared:
+  registry:
+  deployer_bundle:
+
+services:
+  shared:
+    build:
+      context: docker/shared
+    volumes:
+      - shared:/shared
+
+  deployer:
+    privileged: true
+    build:
+      context: docker/deployer
+    environment:
+      - TEST_ID=${TEST_ID}
+    volumes:
+      - ../..:/mrsk
+      - shared:/shared
+      - registry:/registry
+      - deployer_bundle:/usr/local/bundle/
+
+  registry:
+    build:
+      context: docker/registry
+    environment:
+      - REGISTRY_HTTP_ADDR=0.0.0.0:4443
+      - REGISTRY_HTTP_TLS_CERTIFICATE=/certs/domain.crt
+      - REGISTRY_HTTP_TLS_KEY=/certs/domain.key
+    volumes:
+      - shared:/shared
+      - registry:/var/lib/registry/
+
+  vm1:
+    privileged: true
+    build:
+      context: docker/vm
+    volumes:
+      - shared:/shared
+
+  vm2:
+    privileged: true
+    build:
+      context: docker/vm
+    volumes:
+      - shared:/shared
+
+  load_balancer:
+    build:
+      context: docker/load_balancer
+    ports:
+      - "12345:80"

test/integration/docker/deployer/Dockerfile

@@ -0,0 +1,29 @@
+FROM ruby:3.2
+
+WORKDIR /app
+
+RUN apt-get update --fix-missing && apt-get install -y ca-certificates openssh-client curl gnupg docker.io
+
+RUN install -m 0755 -d /etc/apt/keyrings
+RUN curl -fsSL https://download.docker.com/linux/debian/gpg | gpg --dearmor -o /etc/apt/keyrings/docker.gpg
+RUN chmod a+r /etc/apt/keyrings/docker.gpg
+RUN echo \
+  "deb [arch="$(dpkg --print-architecture)" signed-by=/etc/apt/keyrings/docker.gpg] https://download.docker.com/linux/debian \
+  "$(. /etc/os-release && echo "$VERSION_CODENAME")" stable" | \
+  tee /etc/apt/sources.list.d/docker.list > /dev/null
+
+RUN apt-get update --fix-missing && apt-get install -y docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin
+
+COPY *.sh .
+COPY app/ .
+
+RUN ln -s /shared/ssh /root/.ssh
+RUN mkdir -p /etc/docker/certs.d/registry:4443 && ln -s /shared/certs/domain.crt /etc/docker/certs.d/registry:4443/ca.crt
+
+RUN git config --global user.email "deployer@example.com"
+RUN git config --global user.name "Deployer"
+RUN git init && git add . && git commit -am "Initial version"
+
+HEALTHCHECK --interval=1s CMD pgrep sleep
+
+CMD ["./boot.sh"]

test/integration/docker/deployer/app/.env.erb

@@ -0,0 +1 @@
+SECRET_TOKEN=1234

test/integration/docker/deployer/app/.mrsk/hooks/post-deploy

@@ -0,0 +1,3 @@
+#!/bin/sh
+echo "Deployed!"
+mkdir -p /tmp/${TEST_ID} && touch /tmp/${TEST_ID}/post-deploy

test/integration/docker/deployer/app/.mrsk/hooks/pre-build

@@ -0,0 +1,3 @@
+#!/bin/sh
+echo "About to build and push..."
+mkdir -p /tmp/${TEST_ID} && touch /tmp/${TEST_ID}/pre-build

test/integration/docker/deployer/app/.mrsk/hooks/pre-connect

@@ -0,0 +1,8 @@
+#!/bin/sh
+
+echo "About to lock..."
+if [ "$MRSK_HOSTS" != "vm1,vm2" ]; then
+  echo "Expected hosts to be 'vm1,vm2', got $MRSK_HOSTS"
+  exit 1
+fi
+mkdir -p /tmp/${TEST_ID} && touch /tmp/${TEST_ID}/pre-connect

test/integration/docker/deployer/app/.mrsk/hooks/pre-deploy

@@ -0,0 +1,3 @@
+#!/bin/sh
+echo "Deployed!"
+mkdir -p /tmp/${TEST_ID} && touch /tmp/${TEST_ID}/pre-deploy

test/integration/docker/deployer/app/Dockerfile

@@ -0,0 +1,7 @@
+FROM registry:4443/nginx:1-alpine-slim
+
+COPY default.conf /etc/nginx/conf.d/default.conf
+
+ARG COMMIT_SHA
+RUN echo $COMMIT_SHA > /usr/share/nginx/html/version
+

test/integration/docker/deployer/app/config/deploy.yml

@@ -0,0 +1,27 @@
+service: app
+image: app
+servers:
+  - vm1
+  - vm2
+registry:
+  server: registry:4443
+  username: root
+  password: root
+builder:
+  multiarch: false
+  args:
+    COMMIT_SHA: <%= `git rev-parse HEAD` %>
+healthcheck:
+  cmd: wget -qO- http://localhost > /dev/null
+traefik:
+  args:
+    accesslog: true
+    accesslog.format: json
+  image: registry:4443/traefik:v2.9
+accessories:
+  busybox:
+    image: registry:4443/busybox:1.36.0
+    cmd: sh -c 'echo "Starting busybox..."; trap exit term; while true; do sleep 1; done'
+    roles:
+      - web
+stop_wait_time: 1

test/integration/docker/deployer/app/default.conf

@@ -0,0 +1,17 @@
+server {
+    listen       80;
+    listen  [::]:80;
+    server_name  localhost;
+
+    location / {
+        root   /usr/share/nginx/html;
+        index  index.html index.htm;
+    }
+
+    # redirect server error pages to the static page /50x.html
+    #
+    error_page   500 502 503 504  /50x.html;
+    location = /50x.html {
+        root   /usr/share/nginx/html;
+    }
+}

test/integration/docker/deployer/boot.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+
+dockerd --max-concurrent-downloads 1 &
+
+exec sleep infinity

test/integration/docker/deployer/setup.sh

@@ -0,0 +1,23 @@
+#!/bin/bash
+
+install_mrsk() {
+  cd /mrsk && gem build mrsk.gemspec -o /tmp/mrsk.gem && gem install /tmp/mrsk.gem
+}
+
+# Push the images to a persistent volume on the registry container
+# This is to work around docker hub rate limits
+push_image_to_registry_4443() {
+  # Check if the image is in the registry without having to pull it
+  if ! stat /registry/docker/registry/v2/repositories/$1/_manifests/tags/$2/current/link > /dev/null; then
+    hub_tag=$1:$2
+    registry_4443_tag=registry:4443/$1:$2
+    docker pull $hub_tag
+    docker tag $hub_tag $registry_4443_tag
+    docker push $registry_4443_tag
+  fi
+}
+
+install_mrsk
+push_image_to_registry_4443 nginx 1-alpine-slim
+push_image_to_registry_4443 traefik v2.9
+push_image_to_registry_4443 busybox 1.36.0

test/integration/docker/deployer/update_app_rev.sh

@@ -0,0 +1,3 @@
+#!/bin/bash
+
+git commit -am 'Update rev' --amend

test/integration/docker/load_balancer/Dockerfile

@@ -0,0 +1,5 @@
+FROM nginx:1-alpine-slim
+
+COPY default.conf /etc/nginx/conf.d/default.conf
+
+HEALTHCHECK --interval=1s CMD pgrep nginx

test/integration/docker/load_balancer/default.conf

@@ -0,0 +1,16 @@
+upstream loadbalancer {
+  server vm1:80;
+  server vm2:80;
+}
+
+server {
+  listen 80;
+
+  location / {
+    proxy_pass http://loadbalancer;
+    proxy_connect_timeout       10;
+    proxy_send_timeout          10;
+    proxy_read_timeout          10;
+    send_timeout                10;
+  }
+}

test/integration/docker/registry/Dockerfile

@@ -0,0 +1,9 @@
+FROM registry
+
+COPY boot.sh .
+
+RUN ln -s /shared/certs /certs
+
+HEALTHCHECK --interval=1s CMD pgrep registry
+
+ENTRYPOINT ["./boot.sh"]

test/integration/docker/registry/boot.sh

@@ -0,0 +1,5 @@
+#!/bin/sh
+
+while [ ! -f /certs/domain.crt ]; do sleep 1; done
+
+exec /entrypoint.sh /etc/docker/registry/config.yml

test/integration/docker/shared/Dockerfile

@@ -0,0 +1,17 @@
+FROM ubuntu:22.10
+
+WORKDIR /work
+
+RUN apt-get update --fix-missing && apt-get -y install openssh-client openssl
+
+RUN mkdir ssh && \
+  ssh-keygen -t rsa -f ssh/id_rsa -N ""
+
+COPY registry-dns.conf .
+COPY boot.sh .
+
+RUN mkdir certs && openssl req -newkey rsa:4096 -nodes -sha256 -keyout certs/domain.key   -x509 -days 365 -out certs/domain.crt   -subj '/CN=registry' -extensions EXT -config registry-dns.conf
+
+HEALTHCHECK --interval=1s CMD pgrep sleep
+
+CMD ["./boot.sh"]

test/integration/docker/shared/boot.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+
+cp -r * /shared
+
+exec sleep infinity

test/integration/docker/shared/registry-dns.conf

@@ -0,0 +1,7 @@
+[dn]
+CN=registry
+[req]
+distinguished_name = dn
+[EXT]
+subjectAltName=DNS:registry
+keyUsage=digitalSignature

test/integration/docker/vm/Dockerfile

@@ -0,0 +1,14 @@
+FROM ubuntu:22.10
+
+WORKDIR /work
+
+RUN apt-get update --fix-missing && apt-get -y install openssh-client openssh-server docker.io
+
+RUN mkdir /root/.ssh && ln -s /shared/ssh/id_rsa.pub /root/.ssh/authorized_keys
+RUN mkdir -p /etc/docker/certs.d/registry:4443 && ln -s /shared/certs/domain.crt /etc/docker/certs.d/registry:4443/ca.crt
+
+COPY boot.sh .
+
+HEALTHCHECK --interval=1s CMD pgrep dockerd
+
+CMD ["./boot.sh"]

test/integration/docker/vm/boot.sh

@@ -0,0 +1,9 @@
+#!/bin/bash
+
+while [ ! -f /root/.ssh/authorized_keys ]; do echo "Waiting for ssh keys"; sleep 1; done
+
+service ssh restart
+
+dockerd --max-concurrent-downloads 1 &
+
+exec sleep infinity

test/integration/integration_test.rb

@@ -0,0 +1,131 @@
+require "net/http"
+require "test_helper"
+
+class IntegrationTest < ActiveSupport::TestCase
+  setup do
+    ENV["TEST_ID"] = SecureRandom.hex
+    docker_compose "up --build -d"
+    wait_for_healthy
+    setup_deployer
+  end
+
+  teardown do
+    docker_compose "down -t 1"
+  end
+
+  private
+    def docker_compose(*commands, capture: false, raise_on_error: true)
+      command = "TEST_ID=#{ENV["TEST_ID"]} docker compose #{commands.join(" ")}"
+      succeeded = false
+      if capture
+        result = stdouted { succeeded = system("cd test/integration && #{command}") }
+      else
+        succeeded = system("cd test/integration && #{command}")
+      end
+
+      raise "Command `#{command}` failed with error code `#{$?}`" if !succeeded && raise_on_error
+      result
+    end
+
+    def deployer_exec(*commands, **options)
+      docker_compose("exec deployer #{commands.join(" ")}", **options)
+    end
+
+    def mrsk(*commands, **options)
+      deployer_exec(:mrsk, *commands, **options)
+    end
+
+    def assert_app_is_down
+      response = app_response
+      debug_response_code(response, "502")
+      assert_equal "502", response.code
+    end
+
+    def assert_app_is_up(version: nil)
+      response = app_response
+      debug_response_code(response, "200")
+      assert_equal "200", response.code
+      assert_app_version(version, response) if version
+    end
+
+    def assert_app_not_found
+      response = app_response
+      debug_response_code(response, "404")
+      assert_equal "404", response.code
+    end
+
+    def wait_for_app_to_be_up(timeout: 20, up_count: 3)
+      timeout_at = Time.now + timeout
+      up_times = 0
+      response = app_response
+      while up_times < up_count && timeout_at > Time.now
+        sleep 0.1
+        up_times += 1 if response.code == "200"
+        response = app_response
+      end
+      assert_equal up_times, up_count
+    end
+
+    def app_response
+      Net::HTTP.get_response(URI.parse("http://localhost:12345/version"))
+    end
+
+    def update_app_rev
+      deployer_exec "./update_app_rev.sh"
+      latest_app_version
+    end
+
+    def latest_app_version
+      deployer_exec("git rev-parse HEAD", capture: true)
+    end
+
+    def assert_app_version(version, response)
+      assert_equal version, response.body.strip
+    end
+
+    def assert_hooks_ran(*hooks)
+      hooks.each do |hook|
+        file = "/tmp/#{ENV["TEST_ID"]}/#{hook}"
+        assert_equal "removed '#{file}'", deployer_exec("rm -v #{file}", capture: true).strip
+      end
+    end
+
+    def assert_200(response)
+      code = response.code
+      if code != "200"
+        puts "Got response code #{code}, here are the traefik logs:"
+        mrsk :traefik, :logs
+        puts "And here are the load balancer logs"
+        docker_compose :logs, :load_balancer
+        puts "Tried to get the response code again and got #{app_response.code}"
+      end
+      assert_equal "200", code
+    end
+
+    def wait_for_healthy(timeout: 20)
+      timeout_at = Time.now + timeout
+      while docker_compose("ps -a | tail -n +2 | grep -v '(healthy)' | wc -l", capture: true) != "0"
+        if timeout_at < Time.now
+          docker_compose("ps -a | tail -n +2 | grep -v '(healthy)'")
+          raise "Container not healthy after #{timeout} seconds" if timeout_at < Time.now
+        end
+        sleep 0.1
+      end
+    end
+
+    def setup_deployer
+      deployer_exec("./setup.sh") unless $DEPLOYER_SETUP
+      $DEPLOYER_SETUP = true
+    end
+
+    def debug_response_code(app_response, expected_code)
+      code = app_response.code
+      if code != expected_code
+        puts "Got response code #{code}, here are the traefik logs:"
+        mrsk :traefik, :logs
+        puts "And here are the load balancer logs"
+        docker_compose :logs, :load_balancer
+        puts "Tried to get the response code again and got #{app_response.code}"
+      end
+    end
+end

test/integration/lock_test.rb

@@ -0,0 +1,18 @@
+require_relative "integration_test"
+
+class LockTest < IntegrationTest
+  test "acquire, release, status" do
+    mrsk :lock, :acquire, "-m 'Integration Tests'"
+
+    status = mrsk :lock, :status, capture: true
+    assert_match /Locked by: Deployer at .*\nVersion: #{latest_app_version}\nMessage: Integration Tests/m, status
+
+    error = mrsk :deploy, capture: true, raise_on_error: false
+    assert_match /Deploy lock found/m, error
+
+    mrsk :lock, :release
+
+    status = mrsk :lock, :status, capture: true
+    assert_match /There is no deploy lock/m, status
+  end
+end

test/integration/main_test.rb

@@ -0,0 +1,59 @@
+require_relative "integration_test"
+
+class MainTest < IntegrationTest
+  test "deploy, redeploy, rollback, details and audit" do
+    first_version = latest_app_version
+
+    assert_app_is_down
+
+    mrsk :deploy
+    assert_app_is_up version: first_version
+    assert_hooks_ran "pre-connect", "pre-build", "pre-deploy", "post-deploy"
+
+    second_version = update_app_rev
+
+    mrsk :redeploy
+    assert_app_is_up version: second_version
+    assert_hooks_ran "pre-connect", "pre-build", "pre-deploy", "post-deploy"
+
+    mrsk :rollback, first_version
+    assert_hooks_ran "pre-connect", "pre-deploy", "post-deploy"
+    assert_app_is_up version: first_version
+
+    details = mrsk :details, capture: true
+    assert_match /Traefik Host: vm1/, details
+    assert_match /Traefik Host: vm2/, details
+    assert_match /App Host: vm1/, details
+    assert_match /App Host: vm2/, details
+    assert_match /traefik:v2.9/, details
+    assert_match /registry:4443\/app:#{first_version}/, details
+
+    audit = mrsk :audit, capture: true
+    assert_match /Booted app version #{first_version}.*Booted app version #{second_version}.*Booted app version #{first_version}.*/m, audit
+  end
+
+  test "envify" do
+    mrsk :envify
+
+    assert_equal "SECRET_TOKEN=1234", deployer_exec("cat .env", capture: true)
+  end
+
+  test "config" do
+    config = YAML.load(mrsk(:config, capture: true))
+    version = latest_app_version
+
+    assert_equal [ "web" ], config[:roles]
+    assert_equal [ "vm1", "vm2" ], config[:hosts]
+    assert_equal "vm1", config[:primary_host]
+    assert_equal version, config[:version]
+    assert_equal "registry:4443/app", config[:repository]
+    assert_equal "registry:4443/app:#{version}", config[:absolute_image]
+    assert_equal "app-#{version}", config[:service_with_version]
+    assert_equal [], config[:env_args]
+    assert_equal [], config[:volume_args]
+    assert_equal({ user: "root", auth_methods: [ "publickey" ] }, config[:ssh_options])
+    assert_equal({ "multiarch" => false, "args" => { "COMMIT_SHA" => version } }, config[:builder])
+    assert_equal [ "--log-opt", "max-size=\"10m\"" ], config[:logging]
+    assert_equal({ "path" => "/up", "port" => 3000, "max_attempts" => 7, "cmd" => "wget -qO- http://localhost > /dev/null" }, config[:healthcheck])
+  end
+end

test/integration/traefik_test.rb

@@ -0,0 +1,36 @@
+require_relative "integration_test"
+
+class TraefikTest < IntegrationTest
+  test "boot, stop, start, restart, logs, remove" do
+    mrsk :traefik, :boot
+    assert_traefik_running
+
+    mrsk :traefik, :stop
+    assert_traefik_not_running
+
+    mrsk :traefik, :start
+    assert_traefik_running
+
+    mrsk :traefik, :restart
+    assert_traefik_running
+
+    logs = mrsk :traefik, :logs, capture: true
+    assert_match /Traefik version [\d.]+ built on/, logs
+
+    mrsk :traefik, :remove
+    assert_traefik_not_running
+  end
+
+  private
+    def assert_traefik_running
+      assert_match /traefik:v2.9   "\/entrypoint.sh/, traefik_details
+    end
+
+    def assert_traefik_not_running
+      refute_match /traefik:v2.9   "\/entrypoint.sh/, traefik_details
+    end
+
+    def traefik_details
+      mrsk :traefik, :details, capture: true
+    end
+end

test/test_helper.rb

@@ -1,6 +1,7 @@
 require "bundler/setup"
 require "active_support/test_case"
 require "active_support/testing/autorun"
+require "active_support/testing/stream"
 require "debug"
 require "mocha/minitest" # using #stubs that can alter returns
 require "minitest/autorun" # using #stub that take args
@@ -23,4 +24,14 @@ module SSHKit
 end
 
 class ActiveSupport::TestCase
+  include ActiveSupport::Testing::Stream
+
+  private
+    def stdouted
+      capture(:stdout) { yield }.strip
+    end
+
+    def stderred
+      capture(:stderr) { yield }.strip
+    end
 end

test/utils_test.rb

@@ -8,13 +8,16 @@ class UtilsTest < ActiveSupport::TestCase
 
   test "argumentize with redacted" do
     assert_kind_of SSHKit::Redaction, \
-      Mrsk::Utils.argumentize("--label", { foo: "bar" }, redacted: true).last
+      Mrsk::Utils.argumentize("--label", { foo: "bar" }, sensitive: true).last
   end
 
   test "argumentize_env_with_secrets" do
     ENV.expects(:fetch).with("FOO").returns("secret")
-    assert_equal [ "-e", "FOO=\"secret\"", "-e", "BAZ=\"qux\"" ], \
-      Mrsk::Utils.argumentize_env_with_secrets({ "secret" => [ "FOO" ], "clear" => { BAZ: "qux" } })
+
+    args = Mrsk::Utils.argumentize_env_with_secrets({ "secret" => [ "FOO" ], "clear" => { BAZ: "qux" } })
+
+    assert_equal [ "-e", "FOO=[REDACTED]", "-e", "BAZ=\"qux\"" ], Mrsk::Utils.redacted(args)
+    assert_equal [ "-e", "FOO=\"secret\"", "-e", "BAZ=\"qux\"" ], Mrsk::Utils.unredacted(args)
   end
 
   test "optionize" do
@@ -27,13 +30,35 @@ class UtilsTest < ActiveSupport::TestCase
       Mrsk::Utils.optionize({ foo: "bar", baz: "qux", quux: true }, with: "=")
   end
 
-  test "redact" do
-    assert_kind_of SSHKit::Redaction, Mrsk::Utils.redact("secret")
-    assert_equal "secret", Mrsk::Utils.redact("secret")
+  test "no redaction from #to_s" do
+    assert_equal "secret", Mrsk::Utils.sensitive("secret").to_s
+  end
+
+  test "redact from #inspect" do
+    assert_equal "[REDACTED]".inspect, Mrsk::Utils.sensitive("secret").inspect
+  end
+
+  test "redact from SSHKit output" do
+    assert_kind_of SSHKit::Redaction, Mrsk::Utils.sensitive("secret")
+  end
+
+  test "redact from YAML output" do
+    assert_equal "--- ! '[REDACTED]'\n", YAML.dump(Mrsk::Utils.sensitive("secret"))
   end
 
   test "escape_shell_value" do
     assert_equal "\"foo\"", Mrsk::Utils.escape_shell_value("foo")
     assert_equal "\"\\`foo\\`\"", Mrsk::Utils.escape_shell_value("`foo`")
+
+    assert_equal "\"${PWD}\"", Mrsk::Utils.escape_shell_value("${PWD}")
+    assert_equal "\"${cat /etc/hostname}\"", Mrsk::Utils.escape_shell_value("${cat /etc/hostname}")
+    assert_equal "\"\\${PWD]\"", Mrsk::Utils.escape_shell_value("${PWD]")
+    assert_equal "\"\\$(PWD)\"", Mrsk::Utils.escape_shell_value("$(PWD)")
+    assert_equal "\"\\$PWD\"", Mrsk::Utils.escape_shell_value("$PWD")
+
+    assert_equal "\"^(https?://)www.example.com/(.*)\\$\"",
+      Mrsk::Utils.escape_shell_value("^(https?://)www.example.com/(.*)$")
+    assert_equal "\"https://example.com/\\$2\"",
+      Mrsk::Utils.escape_shell_value("https://example.com/$2")
   end
 end