github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 08:30:05 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #4847] Ollama pull model without internet when run with docker #3066

New Issue
Closed
opened 2026-04-12 13:30:11 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-12 13:30:11 -05:00
Owner
Copy Link

Originally created by @david101-hunter on GitHub (Jun 6, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/4847

When I run Ollama run gemma:2b inside docker, it will download blobs folder and manifests folder in /root/.ollama/models.

In private enviroment without interner, I copy blobs & manifests folders, when I run ollama run <name model>, it not working.

How can I achieve this?

Originally created by @david101-hunter on GitHub (Jun 6, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/4847 When I run `Ollama run gemma:2b` inside docker, it will download blobs folder and manifests folder in `/root/.ollama/models`. In private enviroment without interner, I copy `blobs` & `manifests` folders, when I run `ollama run <name model>`, it not working. How can I achieve this?
GiteaMirror added the model label 2026-04-12 13:30:11 -05:00
GiteaMirror commented 2026-04-12 13:30:13 -05:00
Author
Owner
Copy Link

@sealad886 commented on GitHub (Jun 11, 2024):

The Ollama cache is really difficult to manually move because the manifest files for each model implement a symlink-like reference to the underlying blob files that is different from OS-level symlinking; moving the files or changing any of the parent directories' names will break the cache completely.

Note that my solution assumes that your Docker installation binds /root to your $HOME directory. If your virtual storage is configured differently, you may have to tweak this a bit.

To accomplish what you're asking, you would need to first (prior to downloading models) create your own symlink at $HOME/.ollama/models that points to a directory that you intend to use for your downloaded models. For example:

mkdir -p "$HOME/.ollama/sandbox/models"
ln -s -F "$HOME/.ollama/sandbox/models" "$HOME/.ollama/models"

Ollama will still look for models at $HOME/.ollama/models (and that symlink pointer will need to persist), but you could conceivably use this technique with your own setup; tweaks may be required depending on what you're doing with your environments.

Alternatively, you can set the OLLAMA_MODELS environment variable to point to a different download directory. Again, you can't move the cache files or change the names of parent directories, but you could set this to a directory that has shared access with your internet-accessing environment.

<!-- gh-comment-id:2159731000 --> @sealad886 commented on GitHub (Jun 11, 2024): The Ollama cache is really difficult to manually move because the manifest files for each model implement a symlink-like reference to the underlying blob files that is different from OS-level symlinking; moving the files or changing any of the parent directories' names will break the cache completely. Note that my solution assumes that your Docker installation binds `/root` to your `$HOME` directory. If your virtual storage is configured differently, you may have to tweak this a bit. To accomplish what you're asking, you would need to first (prior to downloading models) create your own symlink at `$HOME/.ollama/models` that points to a directory that you intend to use for your downloaded models. For example: ```bash mkdir -p "$HOME/.ollama/sandbox/models" ln -s -F "$HOME/.ollama/sandbox/models" "$HOME/.ollama/models" ``` Ollama will still look for models at `$HOME/.ollama/models` (and that symlink pointer will need to persist), but you could conceivably use this technique with your own setup; tweaks may be required depending on what you're doing with your environments. Alternatively, you can set the `OLLAMA_MODELS` environment variable to point to a different download directory. Again, you can't move the cache files or change the names of parent directories, but you could set this to a directory that has shared access with your internet-accessing environment.
GiteaMirror commented 2026-04-12 13:30:14 -05:00
Author
Owner
Copy Link

@david101-hunter commented on GitHub (Oct 29, 2024):

thanks

<!-- gh-comment-id:2444388072 --> @david101-hunter commented on GitHub (Oct 29, 2024): thanks
GiteaMirror commented 2026-04-12 13:30:15 -05:00
Author
Owner
Copy Link

@mchiang0610 commented on GitHub (Nov 21, 2024):

Hey, for using Ollama offline, you can import the models privately (or pull in advance of disconnecting from the internet)

https://github.com/ollama/ollama/blob/main/docs/import.md

Let me know if this helps! Thank you so much!

<!-- gh-comment-id:2490225630 --> @mchiang0610 commented on GitHub (Nov 21, 2024): Hey, for using Ollama offline, you can import the models privately (or pull in advance of disconnecting from the internet) https://github.com/ollama/ollama/blob/main/docs/import.md Let me know if this helps! Thank you so much!
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 model
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#3066
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?