A few years ago I had a problem. When I was working on iOverlander I realized that there wasn’t an automated PostgreSQL database backup. This was a problem because the database changes very frequently. Thousands of users are creating new Points of Interest and upload check-ins daily and they are all stored in a PostgreSQL database. This is the core of iOverlander, thus having a backup is kinda important. I created a small script that takes a backup. But now the question is where do you store the backup? Of course, you could store it on an S3 compatible storage but then you have to set this up and pay for the storage you use (one database backup was about 800MiB). But maybe there’s a simpler and free way to store a backup at a safe place? In comes Docker Hub! I already pay for a Docker Hub account to have private repository. As far as I know there are no restrictions on how many images / repositories you can have and as long as your image is smaller than 10GiB you shouldn’t run into any problems. So why not wrap the backup in a Docker image and push it to the private repository?! I hacked together a shell script that uses pg_dump to create backup file, creates a Docker image and pushes it to Docker Hub. So far, so good. The next problem is to actually use this backup and restore the database. Since the backup file is in a layer of the Docker image, it’s a bit more complex and you first need to find the right layer and then extract the file. I never came around to write a script for that (and we never had to restore our production database). But whenever another volunteer needed to restore dump locally, I usually had to send them the file or explain the different steps needed to extract the backup from the image. The other day I messed up my local iOverlander database and I had to restore it. It’s been months since I had done this and I didn’t quite remember all the steps. Even though the documentation I wrote wasn’t too bad, it still took me way too long to restore the database. And so I decided to ditch the Shell script and write a CLI tool in Go to encode the whole process and so “BockeR – Backup and Restore in Docker” was born. I don’t think this is gonna be useful for a lot of people, but it is a fun project which helped me learn writting Go code. I have many ideas for improvements and might implement some of them in the future (for example, use the Docker SDK instead of the docker cli). If you are interested you can install bocker via my Homebrew: brew install bueti/tap/bocker Or brew tap bueti/tap brew install bocker or download the binary on Github: https://github.com/bueti/bocker/releases How does it work? Pretty simple: $ bocker --help Bocker is a command line tool which creates a backup from a PostgreSQL database, wraps it in a Docker image, and uploads it to Docker Hub. Of course, Bocker will also do the reverse and restore your database from a backup in Docker Hub. Usage: bocker [command] Available Commands: backup Backup a Postgresql Database completion Generate the autocompletion script for the specified shell help Help about any command restore Restores a Posgres database Flags: -c, --container-id string ID of container running PostgreSQL -d, --db-name string Database name -h, --help help for bocker -n, --namespace string Docker Namespace (default "bueti") -r, --repository string Docker Repository Use "bocker [command] --help" for more information about a command. Example Backup: [email protected]:~$ ./bocker backup -c 491d8559477c -s ioverlander_development -r ioverlander_backup -u ioverlander INFO 2023/02/05 15:59:41 Creating backup... INFO 2023/02/05 16:01:22 Building image... INFO 2023/02/05 16:01:43 Pushing image... Published image bueti/ioverlander_backup:2023-02-05_15-59-41 Example Restore: $ bocker restore -c a16952e2d855 -o ioverlander -t ioverlander_development \ -s ioverlander_development -r ioverlander_backup --tag 2023-02-05_15-59-41 INFO 2023/02/05 17:14:26 Pulling image (bueti/ioverlander_backup:2023-02-05_15-59-41) from registry... INFO 2023/02/05 17:14:28 Extracting backup from Docker image... INFO 2023/02/05 17:14:43 Creating database... INFO 2023/02/05 17:14:44 Database already exists, skipping creation... INFO 2023/02/05 17:14:53 Restoring database... INFO 2023/02/05 17:16:41 Some errors during restore where ignored. Database successfully restored.
Store unfinished entries as drafts!
I built a services to explain tech jargon
This post explains the adjustments that were needed to the way we work in the SRE Team at Ricardo when we went fully remote.
How traveling the Americas shaped my views of Ops, SRE & the World.
Constructing proper PromQL queries for you data is hard. One tool that helps a lot to understand the data that is available and to construct queries is PromLens. If you don’t know PromLens yet, I suggest to read this blog post. A few days ago we open-sourced the Helm Chart we use at Ricardo to deploy PromLens and we’ve made it available on Artifacthub. I hope this Chart will be as useful to you as it is to us.