Unable to run an image if download previously failed due to system running out of space

[directmusic]

Hello,

While attempting to run the command below on my M1 Max, it failed due to an out of disk error.
lume run macos-sequoia-vanilla:latest

ERROR: Failed to pull image error=The operation couldn’t be completed. No space left on device

Upon clearing enough space, running the command again resulted in the following output:

[2025-02-02T17:38:56Z] INFO: Pulling image image=macos-sequoia-vanilla:latest organization=trycua name=macos-sequoia-vanilla_latest registry=ghcr.io
[2025-02-02T17:38:56Z] ERROR: Failed to pull image error=Virtual machine already exists with name: macos-sequoia-vanilla_latest
Error: Virtual machine already exists with name: macos-sequoia-vanilla_latest

When attempting to delete the VM using the command below it fails:

lume delete macos-sequoia-vanilla:latest
Are you sure you want to delete the virtual machine 'macos-sequoia-vanilla:latest'? [y/N] y
[2025-02-02T17:39:47Z] INFO: Deleting VM name=macos-sequoia-vanilla:latest
[2025-02-02T17:39:47Z] ERROR: Failed to delete VM error=Virtual machine not found: macos-sequoia-vanilla:latest
Error: Virtual machine not found: macos-sequoia-vanilla:latest

Running lume prune did not solve the issue and lume ls shows no VMs.

Steps to reproduce the issue:

Run the following command on a system without enough space:
lume run macos-sequoia-vanilla:latest

Expected vs actual behavior

Expected: Running a prebuilt VM again after a failed download should not prevent me from running the same image again.

Actual: Unable to run, delete, or prune the VM after failure.

Your environment (macOS version, lume version)
lume - 0.1.4
macOS Senoma 14.6.1

[f-trycua]

Hi @directmusic - Thanks for the detailed explanation!

I've created these 2 items to keep track of this issue:

• Failed pull should remove the temporary VM dir after reconstructing the image #4
• Delete should still use the remote image name shorthand #5

As a workaround while I roll out a fix, I recommend cleaning up any failed VM installation manually from the ~/.lume folder.

[unitof]

Ditto if download is interrupted (including via ctrl+c).

Can anyone suggest a workaround to manually delete unfinished images? I cannot figure out any way.

~ % lume run macos-sequoia-vanilla:latest
[2025-02-10T23:49:46Z] INFO: Running VM no_display=false shared_directories= name=macos-sequoia-vanilla_latest mount=none
[2025-02-10T23:49:46Z] INFO: Pulling image name=default image=macos-sequoia-vanilla:latest registry=ghcr.io organization=trycua
[2025-02-10T23:49:46Z] ERROR: Failed to pull image error=Virtual machine already exists with name: macos-sequoia-vanilla_latest
[2025-02-10T23:49:46Z] ERROR: Failed to run VM error=Virtual machine already exists with name: macos-sequoia-vanilla_latest
Error: Virtual machine already exists with name: macos-sequoia-vanilla_latest

~ % lume run macos-sequoia-vanilla_latest
[2025-02-10T23:50:07Z] INFO: Running VM name=macos-sequoia-vanilla_latest no_display=false shared_directories= mount=none
[2025-02-10T23:50:07Z] ERROR: Failed to run VM error=Virtual machine not found: macos-sequoia-vanilla_latest
Error: Virtual machine not found: macos-sequoia-vanilla_latest

~ % lume delete macos-sequoia-vanilla_latest
Are you sure you want to delete the virtual machine 'macos-sequoia-vanilla_latest'? [y/N] y
[2025-02-10T23:50:12Z] INFO: Deleting VM name=macos-sequoia-vanilla_latest
[2025-02-10T23:50:12Z] ERROR: Failed to delete VM error=Virtual machine not found: macos-sequoia-vanilla_latest
Error: Virtual machine not found: macos-sequoia-vanilla_latest

[f-trycua]

Hi @unitof - Any corrupted VMs will still be in the ~/.lume folder. You'll need to manually remove any folders with a random GUID. I'll prioritize rolling out a fix for this in the next release.