github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #12093] Add dry_run option to /api/pull endpoint to pre-check feasibility (disk space > model size) #8035

New Issue
Open
opened 2026-04-12 20:16:48 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-12 20:16:48 -05:00
Owner
Copy Link

Originally created by @charlesatzzish on GitHub (Aug 26, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/12093

Motivation

Currently, when running ollama pull <model> via the API, the daemon begins downloading immediately. If there isn’t enough disk space, the process fails after partially downloading potentially several gigabytes.

This has two major downsides:

  1. Poor user experience: wasted bandwidth, wasted time, and late error messages.
  2. System stability risks: if Ollama fills a disk to 100%, the OS can become unstable. For example:
    • Logs and temporary files can’t be written.
    • Databases or config files risk corruption if writes are interrupted.
    • On macOS, Finder/Dock and other services can hang when free space hits zero.
    • In extreme cases, the computer may freeze or crash and fail to reboot.
    • On macOS, rm may fail to run if no disk space available making it impossible to delete the partial download.

Proposal

Introduce a dry_run (boolean) parameter to the existing /api/pull endpoint:

POST /api/pull
{
  "model": "llama3.2:3b-instruct",
  "dry_run": true
}

Instead of downloading the model, Ollama would:

  1. Look up the model manifest in the registry to compute total required size.
  2. Check free disk space in the models directory (~/.ollama/models or $OLLAMA_MODELS).
  3. Return a JSON response indicating whether the pull could succeed without filling the disk.

Example Responses

If sufficient space:

{
  "can_pull": true,
  "needed_bytes": 8350214144,
  "free_bytes": 21474836480
}

If insufficient space:

{

  "can_pull": false,
  "reason": "not enough disk space",
  "needed_bytes": 8350214144,
  "free_bytes": 4294967296
}

Advantages

  1. Protects system stability: avoids scenarios where a pull fills the entire disk and destabilizes the OS.
  2. Improves UX: fail fast before wasting multi-GB downloads.
  3. Minimal API surface: no new endpoint; just an optional parameter on /api/pull.
  4. Extensible: in future, dry_run could also check other requirements (e.g. network, registry reachability).
  5. Safe: read-only operation, doesn’t expose arbitrary system details.

Alternatives considered

  1. A standalone /api/system/disk endpoint → rejected as too broad in scope.
  2. /api/pull gives an error if there is not enough free disk space before starting download → changes current functionality, may break existing apps.

  3. App implements its own file system checks → not possible with a simple web ui that has no file system access
  4. Doing nothing → leaves users vulnerable to wasted downloads and potential severe and unrecoverable system crashes.

Request for feedback

Would maintainers accept a PR that adds this dry_run option to /api/pull?

I’d be happy to implement it and provide tests/documentation.

Thank you for your kind consideration.

Originally created by @charlesatzzish on GitHub (Aug 26, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/12093 ### Motivation Currently, when running `ollama pull <model>` via the API, the daemon begins downloading immediately. If there isn’t enough disk space, the process fails *after* partially downloading potentially several gigabytes. This has two major downsides: 1. **Poor user experience**: wasted bandwidth, wasted time, and late error messages. 2. **System stability risks**: if Ollama fills a disk to 100%, the OS can become unstable. For example: - Logs and temporary files can’t be written. - Databases or config files risk corruption if writes are interrupted. - On macOS, Finder/Dock and other services can hang when free space hits zero. - In extreme cases, the computer may freeze or crash and fail to reboot. - On macOS, rm may fail to run if no disk space available making it impossible to delete the partial download. ### Proposal Introduce a `dry_run` (boolean) parameter to the existing `/api/pull` endpoint: ``` POST /api/pull { "model": "llama3.2:3b-instruct", "dry_run": true } ``` Instead of downloading the model, Ollama would: 1. Look up the model manifest in the registry to compute total required size. 2. Check free disk space in the models directory (~/.ollama/models or $OLLAMA_MODELS). 3. Return a JSON response indicating whether the pull could succeed without filling the disk. Example Responses If sufficient space: ``` { "can_pull": true, "needed_bytes": 8350214144, "free_bytes": 21474836480 } ``` If insufficient space: ``` { "can_pull": false, "reason": "not enough disk space", "needed_bytes": 8350214144, "free_bytes": 4294967296 } ``` ### Advantages 1. Protects system stability: avoids scenarios where a pull fills the entire disk and destabilizes the OS. 2. Improves UX: fail fast before wasting multi-GB downloads. 3. Minimal API surface: no new endpoint; just an optional parameter on /api/pull. 4. Extensible: in future, dry_run could also check other requirements (e.g. network, registry reachability). 5. Safe: read-only operation, doesn’t expose arbitrary system details. ### Alternatives considered 1. A standalone /api/system/disk endpoint → rejected as too broad in scope. 2. /api/pull gives an error if there is not enough free disk space before starting download → changes current functionality, may break existing apps.
 3. App implements its own file system checks → not possible with a simple web ui that has no file system access 4. Doing nothing → leaves users vulnerable to wasted downloads and potential severe and unrecoverable system crashes. ### Request for feedback Would maintainers accept a PR that adds this dry_run option to /api/pull? I’d be happy to implement it and provide tests/documentation. Thank you for your kind consideration.
GiteaMirror added the feature request label 2026-04-12 20:16:48 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label feature request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#8035
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?