macOS showed a very belated warning about the host TiB disk getting full. Hopefully that is the root cause. Planning to re-reinstall Docker Desktop after that, just in case.
After factory restore, it insisted on allocating 99% of my host disk for Docker internals. This soon triggered a macOS prompt about disk space getting low.
Originally, I had that resource pegged to around 200 GB, not 926.35 GB. WTF, Docker.
abuild in Alpine Linux consistently triggers disk space capacity errors in Docker 4.26.0. With Disk usage limit set to 512 GB. On a 1 TiB host mount with 300 GB unused.
I can’t do my work. So pissed that I decided to blindly apply the Docker update. It broke everything, and there’s no way to run my containers anymore.
There are a ridiculous number of problems with this version of Docker Desktop. On top of everything else, factory reset ALWAYS forces the virtual disk to 99% of my 1 TiB host, causing cascading problems for the host OS. FFS.
This seems to be a problem where iTerm2 lacks the Full Disk permission… macOS fails to present a graphical notice about this missing permission… and iTerm2 never prompts to request the permission.
Curious if Tahoe may have altered RBAC in a way that Docker Desktop hasn’t accounted for.
In any case, the official Docker Desktop uninstall feature is silently leaving behind many components, which explains why reinstalling does nothing to fix the virtual disk problems.
Fresh installs are also greedy, attempting to allocate the entire host 1 TiB disk, instead of selecting a reasonable value.
Don’t grab more than 128 GiB by default, please. You end up wrecking the host operating system.
Manually deleted ~/Library/Containers/com.docker.docker, ~/.docker, and ~/Library/Group\ Containers/group.com.docker. Going for my third or fourth Docker Desktop install today.
The stalling behavior was caused by rsubst silently treating an unexported shell variable as blank, which was creating a cp -r… command in the container to copy the system inside itself, filling up disk.