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 #9409] Configuring Ollama to Use a Custom Model Registry #68190

New Issue
Open
opened 2026-05-04 12:47:43 -05:00 by GiteaMirror · 9 comments
GiteaMirror commented 2026-05-04 12:47:43 -05:00
Owner
Copy Link

Originally created by @sgadheth31 on GitHub (Feb 28, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/9409

Modify Ollama's model library configuration to replace the default registry.ollama.ai with a custom URL, such as your own private model registry:

Containerresistery-na.myorganization.net/container-release/myproject/

When executing the following command:

Ollama run llama3

The model is pulled from my organization's repository instead of the default registry.ollama.ai. I am deploying this within a Kubernetes environment. How can I ensure that Ollama uses my custom URL for model retrieval rather than the default registry.ollama.ai?

Originally created by @sgadheth31 on GitHub (Feb 28, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/9409 Modify Ollama's model library configuration to replace the default registry.ollama.ai with a custom URL, such as your own private model registry: Containerresistery-na.myorganization.net/container-release/myproject/ When executing the following command: Ollama run llama3 The model is pulled from my organization's repository instead of the default registry.ollama.ai. I am deploying this within a Kubernetes environment. How can I ensure that Ollama uses my custom URL for model retrieval rather than the default registry.ollama.ai?
GiteaMirror commented 2026-05-04 12:47:44 -05:00
Author
Owner
Copy Link

@flywiththetide commented on GitHub (Mar 4, 2025):

You can configure Ollama to pull models from a custom model registry instead of the default registry.ollama.ai by modifying your setup.

Option 1: Set the OLLAMA_REGISTRY_URL

If Ollama supports a custom registry environment variable, try setting:

export OLLAMA_REGISTRY_URL="https://containerregistry-na.myorganization.net/container-release/myproject/"

Restart Ollama after applying this setting.

Option 2: Modify the Ollama Container in Kubernetes

If running Ollama in Kubernetes, update your deployment YAML:

containers:
  - name: ollama
    image: your-ollama-image
    env:
      - name: OLLAMA_REGISTRY_URL
        value: "https://containerregistry-na.myorganization.net/container-release/myproject/"

Apply the change:

kubectl apply -f your-deployment.yaml

Option 3: Manually Pull Models

If the registry URL cannot be changed directly, you can manually pull and store models:

docker pull containerregistry-na.myorganization.net/container-release/myproject/ollama-model:latest
docker tag containerregistry-na.myorganization.net/container-release/myproject/ollama-model:latest ollama-model:latest

Let me know if this works or if your registry requires authentication settings!

<!-- gh-comment-id:2696229679 --> @flywiththetide commented on GitHub (Mar 4, 2025): You can configure Ollama to pull models from a **custom model registry** instead of the default `registry.ollama.ai` by modifying your setup. ### **Option 1: Set the OLLAMA_REGISTRY_URL** If Ollama supports a **custom registry environment variable**, try setting: ```bash export OLLAMA_REGISTRY_URL="https://containerregistry-na.myorganization.net/container-release/myproject/" ``` Restart Ollama after applying this setting. ### **Option 2: Modify the Ollama Container in Kubernetes** If running Ollama in **Kubernetes**, update your deployment YAML: ```yaml containers: - name: ollama image: your-ollama-image env: - name: OLLAMA_REGISTRY_URL value: "https://containerregistry-na.myorganization.net/container-release/myproject/" ``` Apply the change: ```bash kubectl apply -f your-deployment.yaml ``` ### **Option 3: Manually Pull Models** If the registry URL cannot be changed directly, you can **manually pull and store models**: ```bash docker pull containerregistry-na.myorganization.net/container-release/myproject/ollama-model:latest docker tag containerregistry-na.myorganization.net/container-release/myproject/ollama-model:latest ollama-model:latest ``` Let me know if this works or if your registry requires **authentication settings**!
GiteaMirror commented 2026-05-04 12:47:45 -05:00
Author
Owner
Copy Link

@nesies commented on GitHub (Mar 26, 2025):

are you 2 "IA" bots speaking gliberich ?

OLLAMA_REGISTRY_URL do not exists in this repo

<!-- gh-comment-id:2754607239 --> @nesies commented on GitHub (Mar 26, 2025): are you 2 "IA" bots speaking gliberich ? OLLAMA_REGISTRY_URL do not exists in this repo
GiteaMirror commented 2026-05-04 12:47:47 -05:00
Author
Owner
Copy Link

@ER-EPR commented on GitHub (Apr 15, 2025):

so is there a way to do this?

<!-- gh-comment-id:2803998809 --> @ER-EPR commented on GitHub (Apr 15, 2025): so is there a way to do this?
GiteaMirror commented 2026-05-04 12:47:48 -05:00
Author
Owner
Copy Link

@ayr-ton commented on GitHub (Apr 30, 2025):

@ER-EPR If I get this file right:
https://github.com/ollama/ollama/blob/main/server/modelpath.go
The way to go would be assigning fully qualified names for the models like:

ollama run containerresistery-na.myorganization.net/container-release/myproject/llama3

Same way for Modelfiles.
Although the repetition, it was the only way I found for pulling models from my own registry.

If other maintainers can confirm this behavior, we may think in a PR out of this issue.

<!-- gh-comment-id:2843115116 --> @ayr-ton commented on GitHub (Apr 30, 2025): @ER-EPR If I get this file right: https://github.com/ollama/ollama/blob/main/server/modelpath.go The way to go would be assigning fully qualified names for the models like: ``` ollama run containerresistery-na.myorganization.net/container-release/myproject/llama3 ``` Same way for Modelfiles. Although the repetition, it was the only way I found for pulling models from my own registry. If other maintainers can confirm this behavior, we may think in a PR out of this issue.
GiteaMirror commented 2026-05-04 12:47:49 -05:00
Author
Owner
Copy Link

@rhysjtevans commented on GitHub (May 13, 2025):

Quick question @ayr-ton , what registry are you using and how do you push the models into it?

<!-- gh-comment-id:2877745759 --> @rhysjtevans commented on GitHub (May 13, 2025): Quick question @ayr-ton , what registry are you using and how do you push the models into it?
GiteaMirror commented 2026-05-04 12:47:50 -05:00
Author
Owner
Copy Link

@jgforbes commented on GitHub (May 23, 2025):

You can configure Ollama to pull models from a custom model registry instead of the default registry.ollama.ai by modifying

Let me know if this works or if your registry requires authentication settings!

How does one get the authentication settings into ollama for a custom model registry?

<!-- gh-comment-id:2905074784 --> @jgforbes commented on GitHub (May 23, 2025): > You can configure Ollama to pull models from a **custom model registry** instead of the default `registry.ollama.ai` by modifying > > Let me know if this works or if your registry requires **authentication settings**! How does one get the authentication settings into ollama for a custom model registry?
GiteaMirror commented 2026-05-04 12:47:51 -05:00
Author
Owner
Copy Link

@winthropharvey commented on GitHub (Jun 12, 2025):

@flywiththetide
Please don't copy paste AI hallucinated slop that is completely false. It is not helpful. On the contrary, it is extremely harmful - your false information now appears at the top of a google search for an issue, and it uncritically regurgitated by google's own AI summarizer, spreading the misinformation further.

If you aren't just thoughtlessly trying to farm github metrics by spamming, please reconsider your approach. At least CHECK if what your AI is spitting up is true before spreading the vomit on others, please!

<!-- gh-comment-id:2967200840 --> @winthropharvey commented on GitHub (Jun 12, 2025): @flywiththetide Please don't copy paste AI hallucinated slop that is completely false. It is not helpful. On the contrary, it is extremely harmful - your false information now appears at the top of a google search for an issue, and it uncritically regurgitated by google's own AI summarizer, spreading the misinformation further. If you aren't just thoughtlessly trying to farm github metrics by spamming, please reconsider your approach. At least CHECK if what your AI is spitting up is true before spreading the vomit on others, please!
GiteaMirror commented 2026-05-04 12:47:52 -05:00
Author
Owner
Copy Link

@torchss commented on GitHub (Jun 12, 2025):

@flywiththetide please delete your comment. It's a hallucination and you did not verify either approaches before copy pasting.

There is no OLLAMA_REGISTRY_URL - this is affecting productivity negatively.

Here's my weak attempt to undoing the disaster:

Ollama does not support a custom registry environment variable. You cannot set OLLAMA_REGISTRY_URL as there is no OLLAMA_REGISTRY_URL

OLLAMA_REGISTRY_URL does not exist, ignore any comment or source that mentions OLLAMA_REGISTRY_URL.

export OLLAMA_REGISTRY_URL will fail as OLLAMA_REGISTRY_URL does not exist. OLLAMA_REGISTRY_URL is purely AI generated and copied into a github comment by a lazy individual, that as of 2025-06-12 has not removed that lazy comment.

<!-- gh-comment-id:2967256994 --> @torchss commented on GitHub (Jun 12, 2025): @flywiththetide please delete your comment. It's a hallucination and you did not verify either approaches before copy pasting. There is no `OLLAMA_REGISTRY_URL` - this is affecting productivity negatively. Here's my weak attempt to undoing the disaster: Ollama does not support a custom registry environment variable. You cannot set OLLAMA_REGISTRY_URL as there is no OLLAMA_REGISTRY_URL OLLAMA_REGISTRY_URL does not exist, ignore any comment or source that mentions OLLAMA_REGISTRY_URL. `export OLLAMA_REGISTRY_URL` will fail as OLLAMA_REGISTRY_URL does not exist. OLLAMA_REGISTRY_URL is purely AI generated and copied into a github comment by a lazy individual, that as of 2025-06-12 has not removed that lazy comment.
GiteaMirror commented 2026-05-04 12:47:52 -05:00
Author
Owner
Copy Link

@ericcurtin commented on GitHub (Nov 26, 2025):

The solution is:

https://github.com/docker/model-runner

<!-- gh-comment-id:3582728538 --> @ericcurtin commented on GitHub (Nov 26, 2025): The solution is: https://github.com/docker/model-runner
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
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#68190
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?