docker

Testing out the new Umami demo profile in Drupal 8.6.x.

I wanted to post a quick guide here for the benefit of anyone else just wanting to test out how Lando works or how it integrates with a Drupal project, since the official documentation kind of jumps you around different places and doesn't have any instructions for "Help! I don't already have a working Drupal codebase!":

tl;dr: Go complete the Drupal Local Development Survey, and we'll present the results (among other things) at MidCamp in a couple weeks!

Local development for Drupal is a subject I've invested a lot of time into. At the start of my Drupal journey, I used to use MAMP, then MAMP Pro, then a native *AMP installation. Then when I learned about Vagrant I started building Vagrant-based environments with shell scripts. Then I learned Ansible and started using Vagrant and Ansible. And then I learned Docker and used Ansible, Docker, and sometimes Vagrant!

Everyone's journey is different—but one thing most of us can agree on is: it ain't easy finding a way to run Drupal on your local workstation if you've never done it before.

Should you use MAMP/WAMP/XAMPP? Should you use Acquia Dev Desktop? Should you use Docker or Vagrant and build your own environment? Should you use a packaged solution like Drupal VM or Lando? And then how will you manage your codebase? How will you build a theme?

I recently needed to hack together a setup where Docker containers are spawned by an automated process, then later, a garbage collector runs and kills off all spawned containers. This is on a system where there could be anywhere from tens to hundreds of containers running at any given moment, and I needed traceability of different containers while they're running.

One of the easiest ways to have at-a-glance traceability is to have named containers, e.g. spawned-worker-1, spawned-worker-2, etc.

So if I do a docker ps --format '{{.Names}}', I can get a full list of all running containers by name. And then if I want to filter that list to only show me spawned-worker- prefixed container names, I can pipe the output through grep, awk, and xargs to use the container names in a Docker command, like so:

docker ps --format '{{.Names}}' | grep "^spawned-worker-" | awk '{print $1}' | xargs -I {} docker stop {}

In this case, I'm stopping all containers with spawned-worker- as the start of the name.

I recently had to add support for the MaxMind free GeoIP database to a PHP container build that was based on the official Docker PHP image on Docker Hub. Unfortunately, it seems nobody else who's added this support has documented it, so I figured I'd post this so that the next poor soul who needs to implement the functionality doesn't have to spend half a day doing it!

First, you need the PHP geoip extension, which is available via PECL (note: if you can make the PHP project itself use a composer library, there are a few better/more current geoip libraries available via Packagist!). Here's how to install it in one of the php 5.6 or 7.0-apache images (note that 7.1 uses Debian Stretch instead of Jessie... but the instructions should be the same there):

Since yesterday's announcement that Ansible had released the code behind Ansible Tower, AWX, under an open source license, I've been working on an AWX Ansible role, a demo AWX Vagrant VM, and an AWX Ansible Container project.

As part of that last project, I have published two public Docker Hub images, awx_web and awx_task, which can be used with a docker-compose.yml file to build AWX locally in about as much time as it takes to download the Docker images:

Using the official MySQL Docker image from Docker Hub, I recently ran into the error:

Host '172.18.0.1' is not allowed to connect to this MySQL server

The only change I had made to my docker-compose.yml file was:

mysql:
  image: mysql:5.6
  ports:
    - '3306'
  volumes:
    # Use this option to persist the MySQL DBs in a shared volume.
    - ./mysqldata:/var/lib/mysql:rw,delegated
    # Use this option to persist the MySQL DBs in a data volume.
    # - db_data:/var/lib/mysql

I switched from using a data volume (db_data) to mounting a volume from my host (mysqldata in the current directory), and after the next time I did a docker-compose down and docker-compose up, I started seeing the error about my host not being allowed to connect to the MySQL server.

Recently I had to test importing some very large databases with lots of giant log tables (e.g. 5+ GB tables), and when I tried doing an import into a local docker MySQL container instance, I got ERROR 1114: The table is full. Here are the commands I used:

I found that—likely due to some Docker filesystem defaults—the MySQL import would fail every time when there was a database table containing more than 1GB of data. Now, this could be related to the way the database was exported, and I also found some issues where people were using memory tables that got exported and wouldn't import cleanly.

Drupal VM has been hitting its stride this year; after adding experimental Docker support in 4.5, Docker usage has been refined in 4.6 (with a more stable and useful drupal-vm Docker image, along with a few other things:

Dockerfiles have been able to use ARGs to allow passing in parameters during a docker build using the CLI argument --build-arg for some time. But until recently (Docker's 17.05 release, to be precise), you weren't able to use an ARG to specify all or part of your Dockerfile's mandatory FROM command.

But since the pull request Allow ARG in FROM was merged, you can now specify an image / repository to use at runtime. This is great for flexibility, and as a concrete example, I used this feature to allow me to pull from a private Docker registry when building a Dockerfile in production, or to build from a local Docker image that was created as part of a CI/testing process inside Travis CI.

To use an ARG in your Dockerfile's FROM:

Since I work among a ton of different Linux distros and environments in my day-to-day work, I have a lot of tooling set up that's mostly-OS-agnostic. I found myself in need of a quick barebones CentOS 7 VM to play around in or troubleshoot an issue. Or I needed to run Ubuntu 16.04 and Ubuntu 14.04 side by side and run the same command in each, checking for differences. Or I needed to bring up Fedora. Or Debian.

I used to use my Vagrant boxes for VirtualBox to boot a full VM, then vagrant ssh in. But that took at least 15-20 seconds—assuming I already had the box downloaded on my computer!