Skip to main content

ComfyUI "No space left on device" — out of disk while downloading models

This is a container DISK limit, not VRAM. Model weights download to the container disk, and large checkpoints (a 62GB LTX-2 or 31GB Z-Image) overflow a small volume before the download finishes.

Also seen as: "No space left on device", "disk full", "OSError 28"

Direct answer

How do I fix "Disk Space Exhausted"?

This is a container DISK limit, not VRAM. Model weights download to the container disk, and large checkpoints (a 62GB LTX-2 or 31GB Z-Image) overflow a small volume before the download finishes.

Key facts

  • iContainer/volume disk smaller than the total weight payload.
  • iMultiple large checkpoints (base + refiner + upscaler + LoRAs) downloaded to the same disk.
  • iOld model cache never cleared between deploys.
  • iA note on cloud: persistent-storage flags often do NOT relocate model downloads off the container disk.

Next step

  1. 1Check the real payload size before deploy — sum every .safetensors/.ckpt the workflow references.
  2. 2Provision a bigger container disk sized to the payload plus ~20% headroom.
  3. 3Remove unused checkpoints/refiners from the workflow to shrink the download.
  4. 4On ModelPilot the deploy sizes disk to the model automatically, so heavy models like LTX-2 (62GB) just work.

Not a fit when

  • A 20GB container disk cannot hold a 31GB model no matter how the download retries.

Not sure this workflow will run before you rent a GPU?

Paste your ComfyUI workflow JSON into the free checker — it flags missing nodes, model requirements, and the VRAM/GPU you actually need. No signup.

Check my workflow freeSee GPU pricing

Other ComfyUI errors

Fix

GPU Memory Exhausted

Fix

Model Not Found

Fix

Missing Python Package

Fix

Network Connection Failed

Fix

GPU Driver Error

Fix

Process Terminated

Frequently Asked Questions

What causes "Disk Space Exhausted" in ComfyUI?

Container/volume disk smaller than the total weight payload.

How do I fix "Disk Space Exhausted"?

Check the real payload size before deploy — sum every .safetensors/.ckpt the workflow references. Provision a bigger container disk sized to the payload plus ~20% headroom.