Manual Deploy

Last updated 8 months ago
In your development machine:
Open your production env file with amber encrypt then edit the database URL, and disable or enable logs. Also see Encrypt.
If you have assets and you're using NPM, use npm install . && npm run release to compile/minify .css and .js files.
In your production machine:
Installcrystal and shards commands (See installation guide). You can compile bin/amber tool using shards build amber in your production project to migrate your database.
Then copy your project repository and ensure to setup AMBER_ENCRYPTION_KEY and AMBER_ENV=production in your environments variables. You can get your encryption key from .encryption_key file in your development machine. Then inside your project's folder, install your shards dependencies with shards install --production and compile your executable with shards build <your-app> --release.
Finally run your project executable with bin/<your-app> .
Also you can try to cross-compile to avoid install these dependencies but this feature is a bit buggy on most OS.
​Alpine Linux is great to use crystal build --static flag, though.
Compilation may take a while because --release enables compiler optimizations. You can avoid --release if you server is low end (less 256 RAM)
Amber performance is still acceptable on non-release mode
You may need sudo permission if you're using port 80 or ports < 1000
You can use iptables to redirect port 80 to 8080 or whatever port are you using in production.
iptables -t nat -A PREROUTING -p tcp --dport 80 -j REDIRECT --to-port 8080
Optionally you can setup a system service to monitor your app very easy, and configure a .git/hook/post-receive to deploy like Heroku using git push production . Amber Team is already using a systemd service to manage Amber Framework Website and thepost-receive hook below to manage deployment using git.
#!/bin/bash
set -e
​
export AMBER_ENV=production
​
if [ "$GIT_DIR" = "." ]; then
  # The script has been called as a hook; chdir to the working copy
  cd ..
  unset GIT_DIR
fi
​
# try to obtain the usual system PATH
if [ -f /etc/profile ]; then
  PATH=$(source /etc/profile; echo $PATH)
  export PATH
fi
​
# get the current branch
head="$(git symbolic-ref HEAD)"
​
# read the STDIN to detect if this push changed the current branch
while read oldrev newrev refname
do
  [ "$refname" = "$head" ] && break
done
​
# abort if there's no update, or in case the branch is deleted
if [ -z "${newrev//0}" ]; then
  exit
fi
​
# check out the latest code into the working copy
umask 002
git reset --hard
​
logfile=log/deploy.log
​
if [ -z "${oldrev//0}" ]; then
  # this is the first push; this branch was just created
  mkdir -p log tmp bin
  chmod 0775 log tmp bin
  touch $logfile
  chmod 0664 $logfile
​
  # init submodules
  git submodule update --recursive --init 2>&1 | tee -a $logfile
else
  # log timestamp
  echo ==== $(date) ==== >> $logfile
​
  ################################################################################
  ####  BEGIN: Amberframework specific Code                                  #####
  ################################################################################
​
  shards install
​
  if [ -f bin/amber ]; then
    echo "amber already installed"
  else
    crystal build lib/amber/src/amber/cli.cr -o bin/amber --stats
  fi
​
  echo "Migrating..."
  ./bin/amber db create migrate || true
  echo "Building application in release mode"
  crystal build src/amberframework.cr -o bin/amberframework --release --stats
  systemctl status amberframework
  sudo systemctl restart amberframework
  systemctl status amberframework
fi
Then add your production repository to your development machine using git remote add
git remote add production <user>@<ip>:/home/<user>/<repository>
<user> is a valid user with your public ssh key inside ~/.ssh/authorized_keys
<ip> is your production machine IP (you can use a domain as well).
<repository> is your project with your.git folder inside.
Deployment
Digital Ocean