Firestore backup with GitHub Actions

I needed a way to perform automated backup for my GCP Firestore database. I found out that Firestore has Import/ Export functionality, and it seems to be the recommended way to do backup. So I created a simple GitHub Actions workflow to do this:

name: Firestore backup

on:
  workflow_dispatch:
  schedule:
    - cron: "0 19 * * *"

jobs:
  backup:
    runs-on: ubuntu-latest

    steps:
      - uses: GoogleCloudPlatform/github-actions/[email protected]
        with:
          service_account_email: '<service-account-name>@<project-id>.iam.gserviceaccount.com'
          service_account_key: ${{ secrets.SA_KEY }}
          project_id: '<project-id>'

      - name: Backup
        run: |-
          gcloud components install beta
          gcloud beta firestore export gs://<storage-bucket-id>


This workflow can be triggered in 2 ways, manually (due to workflow_dispatch trigger), or on a schedule (cron syntax).

The first step set up and authenticate the gcloud cli. The second step triggers to export action on GCP.
Currently firestore export is only available on gcloud beta, so an installation step of the beta component is needed.

The cloud storage bucket would need to be created beforehand. I created mine with the NEARLINE storage class to save on some cost.

“Affluence without abundance” – James Suzman

This was the first book I read in 2020, and the first book I’ve actually read from start to finish in many years. I picked it up hoping to hear some philosophical wisdom on human happiness, or social commentary on the state of Western late-stage capitalism and obsessive consumerism culture. In hindsight, perhaps I was seeking for some form of validation of my own personal feelings.

Needless to say, I found none of that in the book. Instead, it was an incredibly rich description of life in Southern Africa. It was a good reminder how humans lived as hunters and gatherers for hundreds of thousand of years where they only take as much as they needed for a few days at a time from their surroundings. While there were lean periods, in general they do not spend more than 15-20 hours a week in the pursuit of sustenance. Land rights are one of the many modern concepts that upended that way of life, even though it has been around for much longer than what we think of “modern” human history, whether you measure that in thousand of years, or hundred of years since the industrial revolution. One could argue that where we are today is an inevitable destination of the human psychological evolution journey, and I certainly don’t have the knowledge or insight to affirm or deny that. Nonetheless, it was interesting to be presented with the incredible research and understanding on how life was and how it has been drastically altered in the course of recent human history.

CI/CD pipeline comparison – Jenkins vs. GitHub Actions

I’ve been asked recently whether I would use Jenkins or GitHub Actions to create a CI pipeline for a web app project. This got me thinking a bit, as it’s a comparison I’ve often thought about but never had to make a decision on.

I’m fully aware of the multitudes of caveats and “it depends” of such a vague premise. There are also multiple other hosted services as well as open source projects to choose from, such as GitLab CI, Circle CI, Azure DevOps, to name a few. There are also stack-specific tools to choose from, such as Netlify, Vercel, or the various tools in the Kubernetes space. It’s unrealistic for me to do a comprehensive comparison, or experience them all.

I just want to note down some thoughts after having some experience using these two systems. YMMV.

Jenkins

If you work in an “enterprise”, or a big team with a variety of different tech stacks, it might be worthwhile to invest the time and effort into maintaining your own Jenkins server, or look into a hosted offering.

It’s a tried and true option, with tons of resources and support, and is infinitely extensible. The new style of Declarative Pipeline has made reading and maintaining Jenkins pipeline more enjoyable, while the ability to embed Groovy script into your pipeline makes it very flexible.

Another benefit is a fairly complex permissions and security model, with support for RBAC and secrets management.

GitHub Actions

While Jenkins has been around forever, GA is the shiny new toy. Its biggest advantage is perhaps its beautiful UI integrated directly into where your code lives. It takes away the burden of maintaining a separate system to test and build your code.

This might be a bit controversial, but I am not too fond of the yaml syntax chosen. While it’s very popular in certain circles, I find it limiting when you want to do some highly customized logic, which I’ve found to be pretty common in a build pipeline.

Similar to Jenkins extensible plugin model, GA also allows you to reuse code, which is pretty cool and powerful. Sometimes trying to incorporate these shared actions can feel a bit awkward.

What to choose

As I started to write down some of these thoughts, I realized that there are just too many factors to consider here. I could write several posts discussing each of these topics in details. But that might not be too useful for anyone.

My general opinion is, if you’re already using GitHub, then spend a couple days building out GitHub Actions workflow. If you can get it to work and do everything you need, consider it a win! I think it will require the least amount of maintenance going forward, and thus will be a higher return on investment.

If you find yourself struggling with it too much, then don’t be afraid to spin up a Jenkins server. There’s a pretty good chance you’ll be able to get it to do exactly what you need.

If you have other thoughts and opinions, feel free to reach out and have a conversation. I spend a good amount of my time building pipelines, and am always looking to learn more.

ievms – Easily bring up IE browsers on any environment with Vagrant and VirtualBox

Thinking about incorporating IE browser testing for your site, but unsure how to proceed given that you’re developing on a macOS or Linux environment? Do you want to bring up a Windows virtual machine with something like VirtualBox, but not sure how to grab a valid Windows license? Luckily, Microsoft is aware of how difficult it is to develop and test for Edge and IE browsers, so they have released free virtual machines that already have these browsers built in.

If you go to the Microsoft developer website above, you have the options of downloading these VM images for different virtualization softwares (for e.g. VirtualBox and VMWare). However, these steps still feel pretty manual. I wanted to have a simple set of instructions that I could pass around to folks on my team to do IE testing with. Ideally, it would just be a script that anyone can run from any development environment. Thankfully, Microsoft also made vagrant images available, so I was able to create a simple wrapper around this.

ievms – a simple way to start IE and Edge VMs with Vagrant and Virtual Box

Vagrant is a developer tool that allows you to create automated and reproducible development environment using virtual machines. It works pretty well with VirtualBox, a free and well-supported virtualizer.

ievms relies on vagrant, VirtualBox and the images provided Microsoft to create a simple workflow for managing IE browsers across different Windows versions.

Bringing up a Windows 10 virtual machine with IE 10 installed and ready to use:

:; vagrant up ie10-win7

This will download the image if it’s not installed, add it to vagrant, and then bring up the virtual machine with a graphical UI. Once you’re done testing, you can suspend the machine with vagrant suspend ie10-win7, or remove it completely with vagrant destroy ie10-win7. Even if you removed it, the next time you need to bring it up, the image is already cached locally, so you will not have to wait for the download again.

In addition to IE10 on Windows 7, other browser-platform combinations are supported by default:
– ie6-xp
– ie8-xp
– ie7-vista
– ie8-win7
– ie9-win7
– ie10-win7
– ie11-win7
– ie10-win8
– ie11-win81
– msedge-win10

Bonus: You can also test a local website that’s running on the host development environment. For example, if you have a site running on port 8080 locally, you can reach it from within the virtual machine by going to http://192.168.33.1:8080.

Check out ievms‘s README for more instruction on how to get started.