DEV Community

Yash Thakkar
Yash Thakkar

Posted on • Updated on

CockroachDB auto-backup with Docker

CockroachDB is one of the most popular cloud-native databases. CockroachDB is an ACID compliant, relational database that’s wire compatible with PostgreSQL. CockroachDB delivers full ACID transactions at scale even in a distributed environment and guarantees serializable isolation in a cloud-neutral distributed database. We can deploy it using Docker and Kubernetes without any issue. CockroachDB delivers on the key cloud-native primitives of horizontal scale, no single points of failure, survivability, automatable operations, and no platform-specific encumbrances.

How to run CockroachDB?

We can run CockroachDB using a single executable binary or using Docker and Kubernetes. We will see how to run CockroachDB in Docker and how to take backup as well. You can follow the blog or Docs and follow the backup instructions.

Start CockroachDB using Docker

We can run a single or multi node cluster with Docker but for development, we will be using single node cluster only.

Below command will pull the latest CockroachDB image from Docker hub, bind 26257, 8080 ports, bind the volume with cockroach_data/roach1. Run the command and go to http://localhost:8080, you will be able to see the CockroachDB Dashboard.

docker run -d \
      --name=roach1 \
      --hostname=roach1 \
      -p 26257:26257 -p 8080:8080  \
      -v "${PWD}/cockroach_data/roach1:/cockroach/cockroach-data"  \
      cockroachdb/cockroach:latest start \
Enter fullscreen mode Exit fullscreen mode

Now, our CockroachDB database server is up and running. Now, Let's create some data using workload init command or you can dump existing database as well. Run below commands to do that. Here is the docs explaining in details what we are doing in below command.

docker exec -it roach1 bash

./cockroach workload init movr 'postgresql://root@localhost:26257?sslmode=disable'
./cockroach sql --insecure --host=localhost:26257
USE movr;
SHOW tables;
Enter fullscreen mode Exit fullscreen mode

As you can understand, it will create a database movr with some tables and records. You should be able to see below output on your terminal.

(6 rows)

Time: 9.579561ms
Enter fullscreen mode Exit fullscreen mode

This means, we have successfully generated demo data and it's time to take a backup of it. Run exit 2 times to get out of the container. 1st exit command to get out of cockroach sql command line and 2nd is to exit from the container.

CockroachDB Database backup

Our database is ready, now it's time to take a backup of it. To do that, there are 2 ways, we can take a backup of it.


CockroachDB's BACKUP statement allows you to create full or incremental backups of your cluster's schema and data that are consistent as of a given timestamp. Backups can be with or without revision history.

There are many advantages of this process. We can setup whether we want to take a full backup or Incremental backup, automate backup with JOBS, upload it to Amazon, Azure, Google Cloud, NFS or any S3-compatible services, backup a single table or view and many more. You can read more on it here.

This happens inside cockroachDB container environment, so cockroachDB has full control over it.

The only disadvantage is this is available only for enterprise users. This means that, if we are running cockroachDB locally or on a small server, where we may not want enterprise support, we can't use this feature. We can CockroachCloud, if we are running it on a small scale and planning to scale it in the future. CockroachCloud provides a Fully hosted and managed, Self-service platform with Enterprise features and basic support. You can read more here.

So how can we do it without BACKUP feature, which is available only for CockroachCloud and CockroachDB Enterprise users.

2. cockroach dump

CockroachDB provides dump command, which is similar to pg_dump. The cockroach dump command outputs the SQL statements required to recreate tables, views, and sequences. This command can be used to back up or export each database in a cluster. The output should also be suitable for importing into other relational databases, with minimal adjustments. You can read more here.

That's great, so now, all we need to do it run a cron that executes cockroach dump whenever we want and that's it.

But, there are many things we have to think about. Like, how and where we are going to store our backup, how to take a backup on-demand etc etc.

Exactly, for that, I have created an open source docker image with the above features. So let's go through it one by one.

  • Customize cron with CRON_SCHEDULE env.
  • Manual backup at any time
  • Optional backup AWS S3/Spaces upload, if you provide ACCESS_KEY_ID, then it will take it as you want backup to uploaded on S3 or Spaces or anywhere compatible with s3 API.
  • All cockroach image env variable support, you can override COCKROACH_USER, COCKROACH_INSECURE etc. docs
  • It exposes /data as volume, which contains backup zip file, so we can use backup from here if we don't want to upload it to any S3 services.
Environment Variables
  • ACCESS_KEY_ID: Spaces access key id
  • BUCKET_NAME: Spaces bucket name
  • S3_URL: AWS S3( or DO Spaces(
  • SECRET_ACCESS_KEY: Spaces secret access key

Now, let's look at how we can run our docker image to take a backup. Below, is the sample, how we can run the Docker container, which will take a backup of our movr database everyday and upload it to S3 service like AWS S3, Digital Ocean Spaces etc.

docker run -d \
      --name cockroach-backup \
      -v $(pwd)/data:/data \
      -v $(pwd)/cockroach-certs:/cockroach-certs \
      -e S3_URL=S3_URL \
      -e COCKROACH_DATABASE=movr \
      -e COCKROACH_HOST=localhost \
      -e CRON_SCHEDULE="0 0 * * *" \
      -e COCKROACH_USER=root
Enter fullscreen mode Exit fullscreen mode

With the above command, we will run our backup container name cockroach-backup with /data volume, which will contain all the backup files and with ACCESS_KEY_ID, we can upload it to wherever we want. It supports every client connection params.

Now, let's say, you want to take a database backup of the current moment, you can run a below command to trigger manual backup as well.

curl -X POST http://localhost:9000/backup
Enter fullscreen mode Exit fullscreen mode

These features are already available and few more already on the list. Feel free to open an issue to add more features.

This is just a docker image so it works with Kubernetes as well.

Upcoming features:

  • Flags support
  • Multiple database backup support

Note: This is an open source project with MIT. This is my first golang project, I am a newbie, so I may have made mistakes. Issues and pull requests are most welcome.


Top comments (0)