Disk full of orphan files? How to clean Docker volumes

Why does Docker consume so much disk space?

Anyone who uses containerization tools on a regular basis has certainly encountered an annoying operating system warning stating that storage has reached its limit. In practice, this means the physical machine is choked by gigabytes of temporary files that have lost their usefulness. Docker was designed to be fast and isolated, but this architecture causes it to store every single download chunk, build cache, and old version history to speed up future runs. The problem is that over the months, these files become orphaned, meaning they continue taking up physical space even when the corresponding application has already been deleted or replaced by a newer version.

To understand how this problem builds up, remember that every container relies on base images downloaded from the internet. When we update an app, Docker downloads a new image and marks the old one as none, commonly known as a dangling image. Furthermore, volumes created to retain important data frequently outlive the containers that spawned them, turning into forgotten islands of data. Before wiping everything out in a panic, it is worth checking complementary best practices to optimize your environment, much like what we discussed about how to clean Docker cache and free up disk space during previous maintenance tasks.

Identifying the size of the problem on your server

Before running any cleanup command, the first sensible step is to find out exactly where storage is being wasted. Docker features a fantastic native tool for this investigation called system df. When executed in the terminal, it displays a clear dashboard dividing storage across images, containers, local volumes, and build cache. In practice, this command acts like a quick X-ray showing the total allocated space and the exact percentage currently in use by active applications.

To run this analysis, simply open your terminal and type the command below with appropriate privileges:

docker system df

The output will display detailed columns with total space and reclaimable space, which represents the exact volume of data that can be safely recovered. If the reclaimable space number is high, it means there is a lot of accumulated bloat that can be eliminated without any risk to running production services. This diagnostic prevents you from making decisions based solely on guesswork, focusing your cleanup efforts precisely where the storage bottleneck is largest.

Removing stopped containers and orphan images

When we shut down a container using standard commands, it usually remains registered in the system with a stopped status. These records store configurations and logs that are rarely consulted again, yet they continue consuming metadata and disk space. To eliminate all containers that are not currently running in one go, we use the specific pruning command for containers. In practice, this action performs a general sweep and clears out obsolete history from all inactive instances in the environment.

The standard command to perform this cleanup of inactive containers is:

docker container prune

The terminal will request security confirmation before proceeding with permanent deletion. Once confirmed, the system will remove all stopped instances and return the corresponding space to the operating system. Next, we can do the same with orphan images that are not linked to any active container by using the image prune command. If you also manage other infrastructure tools on a daily basis, it is worth checking similar procedures to keep your system clean, such as how to clean APT package cache and free disk space on Linux.

Cleaning unused volumes without losing data

Docker volumes are mechanisms designed to persist data generated by and for containers, ensuring important information does not disappear when a container is destroyed. The major issue occurs when a container is permanently removed, but its corresponding volume is forgotten on the disk, turning into an orphan file that silently consumes gigabytes. To resolve this safely, we use the volume cleanup command, which intelligently differentiates between volumes connected to active containers and those completely abandoned.

Run the command below to scan and remove volumes that have no current bindings:

docker volume prune

It is crucial to point out that Docker is smart enough not to delete volumes currently in use by active containers, even if they are momentarily stopped. However, before running this command on production servers, it is always prudent to check whether there are any deactivated containers that need to be restarted later with those specific data sets. If you have doubts about which volumes exist on the machine, you can list all of them beforehand using docker volume ls to inspect names and ensure no critical information is lost during the process.

Running a general cleanup with system prune

If you want to automate the entire house-cleaning process at once without running separate commands for containers, images, and networks, Docker offers a drastic and comprehensive solution called system prune. In practice, this command acts like a complete spring cleaning, sweeping away all inactive system resources in one go. It removes stopped containers, unused networks, untagged images, and all accumulated build cache over time.

To execute a general cleanup with the aggressive mode that also targets unused images, use the additional parameter:

docker system prune -a --volumes

Using the -a flag ensures that all images not currently used by active containers are removed, rather than just untagged ones. Meanwhile, the --volumes parameter ensures that unused volumes are also included in the deletion package. This routine is extremely useful in development environments or test servers where digital clutter accumulates very frequently. In production environments, however, it should be used cautiously to avoid accidentally removing base images that would take a long time to download again. To explore these architectural concepts further in production, see our related technical analysis on How to inspect Kubernetes logs in real time with Stern.

Final considerations

Keeping a Docker environment clean and organized is an essential practice to avoid unpleasant surprises with full disks during critical operational moments. We have seen that native tools like pruning commands make this task quick, allowing you to recover precious storage gigabytes with a few terminal clicks. Integrating these preventive maintenance routines into the infrastructure lifecycle guarantees greater stability and predictability for servers.

Adopting the habit of checking disk consumption regularly and discarding what is no longer useful extends hardware lifespan and simplifies technical daily operations. Whether in homelab servers or robust corporate environments, mindful container and volume management is what separates a chaotic operation from a clean, reliable, and efficient system.