Disk Space Completely Consumed When Using Docker Operator

[k----n]

I've come across an issue where my VM runs out of disk space after SPO has been running for 5 months.

It looks like the /var/lib/docker/overlay2 consumes all the disk space and the redis container can no longer start.

To resolve it, I stopped the shinyproxy operator, cleaned up Docker with docker system prune -a, remove the redis data folder for shinyproxy operator, removed all old containers associated with shinyproxy operator, and restarted shinyproxy operator again.

Is there any way to avoid this?

[LEDfan]

Hi

The images used by the operator (and ShinyProxy itself) are relatively small. Usually the images of the apps (especially Shiny apps) are way bigger, for example between 1 and 2GB. There is currently no cleanup mechanism included in the operator, since it's non-trivial to determine which images are no longer needed.

What could help is to regularly run docker image prune, this will cleanup dangling images. These are images that no longer have a valid tag. In the context of ShinyProxy, this are typically older versions of your images, which are no longer in use.

If you build images on your server directly, it can also help to run docker builder prune. This will cleanup the build cache which can sometimes tens GBs.

I'll keep this issue open as an enhancement, since it's something we would ideally try to fix in the operator, even if it's not simple.