github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 08:59:13 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #4454] OLLAMA_MODELS environment variable is not respected #2783

New Issue
Closed
opened 2026-04-12 13:06:34 -05:00 by GiteaMirror · 6 comments
GiteaMirror commented 2026-04-12 13:06:34 -05:00
Owner
Copy Link

Originally created by @JLCarveth on GitHub (May 15, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/4454

What is the issue?

I have followed the steps here to change where Ollama stores the downloaded models.

I make sure to run systemctl daemon-reload and to restart the ollama service, and yet it is still storing the model blobs in /usr/share/ollama/... instead of the location specified in OLLAMA_MODELS.

I expect Ollama to download the models to the specified location. I have insufficient space left on my root partition, hence why I am trying to download the models to my home directory instead.

sudo systemctl edit ollama.service:

### Anything between here and the comment below will become the contents of the drop-in file

Environment="OLLAMA_MODELS=/home/jlcarveth/.ollama/models"

OS

Linux

GPU

AMD

CPU

Intel

Ollama version

0.1.32

Originally created by @JLCarveth on GitHub (May 15, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/4454 ### What is the issue? I have followed the steps [here](https://github.com/ollama/ollama/blob/main/docs/faq.md#where-are-models-stored) to change where Ollama stores the downloaded models. I make sure to run `systemctl daemon-reload` and to restart the ollama service, and yet it is still storing the model blobs in `/usr/share/ollama/...` instead of the location specified in `OLLAMA_MODELS`. I expect Ollama to download the models to the specified location. I have insufficient space left on my root partition, hence why I am trying to download the models to my home directory instead. `sudo systemctl edit ollama.service`: ``` ### Anything between here and the comment below will become the contents of the drop-in file Environment="OLLAMA_MODELS=/home/jlcarveth/.ollama/models" ``` ### OS Linux ### GPU AMD ### CPU Intel ### Ollama version 0.1.32
GiteaMirror added the bug label 2026-04-12 13:06:34 -05:00
GiteaMirror commented 2026-04-12 13:06:35 -05:00
Author
Owner
Copy Link

@mxyng commented on GitHub (May 15, 2024):

Your drop in files is missing a [Service] header so systemd doesn't know what section the configuration belongs in.

This should be the full contents of the configuration file:

[Service]
Environment="OLLAMA_MODELS=/home/jlcarveth/.ollama/models"

The FAQ example uses OLLAMA_HOST but the idea is the same

<!-- gh-comment-id:2112849697 --> @mxyng commented on GitHub (May 15, 2024): Your drop in files is missing a `[Service]` header so systemd doesn't know what section the configuration belongs in. This should be the full contents of the configuration file: ``` [Service] Environment="OLLAMA_MODELS=/home/jlcarveth/.ollama/models" ``` The FAQ [example](https://github.com/ollama/ollama/blob/main/docs/faq.md#setting-environment-variables-on-linux) uses `OLLAMA_HOST` but the idea is the same
GiteaMirror commented 2026-04-12 13:06:37 -05:00
Author
Owner
Copy Link

@icf20 commented on GitHub (Jun 6, 2024):

@mxyng setting said variable like u posted it prevents ollama from starting

cat /etc/systemd/system/ollama.service.d/override.conf                                                                                                                      
[Service]
Environment="OLLAMA_MODELS=/mnt/media/ollama/models"

systemctl status ollama                                                                                                                                                   ✔ 
● ollama.service - Ollama Service
     Loaded: loaded (/usr/lib/systemd/system/ollama.service; disabled; preset: disabled)
    Drop-In: /etc/systemd/system/ollama.service.d
             └─override.conf
     Active: activating (auto-restart) (Result: exit-code) since Thu 2024-06-06 13:53:45 CEST; 2s ago
    Process: 23510 ExecStart=/usr/bin/ollama serve (code=exited, status=1/FAILURE)
   Main PID: 23510 (code=exited, status=1/FAILURE)
        CPU: 26ms
<!-- gh-comment-id:2152210991 --> @icf20 commented on GitHub (Jun 6, 2024): @mxyng setting said variable like u posted it prevents ollama from starting ``` cat /etc/systemd/system/ollama.service.d/override.conf [Service] Environment="OLLAMA_MODELS=/mnt/media/ollama/models" ``` ``` systemctl status ollama  ✔ ● ollama.service - Ollama Service Loaded: loaded (/usr/lib/systemd/system/ollama.service; disabled; preset: disabled) Drop-In: /etc/systemd/system/ollama.service.d └─override.conf Active: activating (auto-restart) (Result: exit-code) since Thu 2024-06-06 13:53:45 CEST; 2s ago Process: 23510 ExecStart=/usr/bin/ollama serve (code=exited, status=1/FAILURE) Main PID: 23510 (code=exited, status=1/FAILURE) CPU: 26ms ```
GiteaMirror commented 2026-04-12 13:06:37 -05:00
Author
Owner
Copy Link

@icf20 commented on GitHub (Jun 6, 2024):

for anyone reading or future me

had to copy usr/lib/systemd/system/ollama.service and edit that

### Anything between here and the comment below will become the contents of the drop-in file

[Unit]
Description=Ollama Service
Wants=network-online.target
After=network.target network-online.target

[Service]
#ExecStart=/usr/bin/ollama serve
#WorkingDirectory=/var/lib/ollama
Environment="GIN_MODE=release" "HOME=/mnt/media/ollama" "OLLAMA_MODELS=/mnt/media/ollama/models"
User=user
Group=user
Restart=on-failure
RestartSec=3
Type=simple
PrivateTmp=yes
ProtectSystem=full
ProtectHome=yes
<!-- gh-comment-id:2152748370 --> @icf20 commented on GitHub (Jun 6, 2024): for anyone reading or future me had to copy usr/lib/systemd/system/ollama.service and edit that ``` ### Anything between here and the comment below will become the contents of the drop-in file [Unit] Description=Ollama Service Wants=network-online.target After=network.target network-online.target [Service] #ExecStart=/usr/bin/ollama serve #WorkingDirectory=/var/lib/ollama Environment="GIN_MODE=release" "HOME=/mnt/media/ollama" "OLLAMA_MODELS=/mnt/media/ollama/models" User=user Group=user Restart=on-failure RestartSec=3 Type=simple PrivateTmp=yes ProtectSystem=full ProtectHome=yes ```
GiteaMirror commented 2026-04-12 13:06:37 -05:00
Author
Owner
Copy Link

@xnming commented on GitHub (Jan 26, 2025):

After researching online and consulting with ChatGPT, I was able to resolve the issue successfully. For anyone facing a similar problem on Linux, ensure that the ollama user has write permissions for the specified path. Below is a complete tutorial on setting up OLLAMA_MODELS:

1. Identify the User Running the Ollama Service

To ensure the correct permissions are applied, determine the username of the account running the Ollama service.

ps aux | grep ollama    # in most cases, the username would be 'ollama'

The username will appear in the first column of the output.

2. Create a New Folder for Models

Choose or create a directory for storing the models

mkdir -p /path/to/ollama/models

Replace /path/to/ollama/models with your desired directory path.

3. Change Ownership and Permissions

Grant ownership and full write permissions of the directory to the user running the Ollama service.

sudo chown -R ollama_user /path/to/ollama/models
sudo chmod -R 755 /path/to/ollama/models

Replace ollama_user with the username obtained in Step 1.

4. Edit the Ollama Service Configuration

Update the Ollama service configuration to include the OLLAMA_MODELS environment variable.

  1. Open the service file
sudo systemctl edit ollama.service
  1. Add the following lines under the [Service] section:
[Service]
Environment="OLLAMA_MODELS=/path/to/ollama/models"
  1. Save and exit the editor.

5. Reload and Restart the Ollama Service

Apply the changes by reloading the system configuration and restarting the service.

sudo systemctl daemon-reload
sudo systemctl restart ollama

6. Verify the Configuration

Ensure the service is running without errors:

sudo systemctl status ollama

Test the connectivity:

curl http://localhost:11434/api/version

If the service responds with the version details, the configuration is successful.

<!-- gh-comment-id:2614231357 --> @xnming commented on GitHub (Jan 26, 2025): After researching online and consulting with ChatGPT, I was able to resolve the issue successfully. For anyone facing a similar problem on Linux, ensure that the `ollama` user has write permissions for the specified path. Below is a complete tutorial on setting up `OLLAMA_MODELS`: ### 1. Identify the User Running the Ollama Service To ensure the correct permissions are applied, determine the username of the account running the Ollama service. ```bash ps aux | grep ollama # in most cases, the username would be 'ollama' ``` The username will appear in the first column of the output. ### 2. Create a New Folder for Models Choose or create a directory for storing the models ```bash mkdir -p /path/to/ollama/models ``` Replace /path/to/ollama/models with your desired directory path. ### 3. Change Ownership and Permissions Grant ownership and full write permissions of the directory to the user running the Ollama service. ```bash sudo chown -R ollama_user /path/to/ollama/models sudo chmod -R 755 /path/to/ollama/models ``` Replace ollama_user with the username obtained in Step 1. ### 4. Edit the Ollama Service Configuration Update the Ollama service configuration to include the OLLAMA_MODELS environment variable. 1. Open the service file ```bash sudo systemctl edit ollama.service ``` 2. Add the following lines under the `[Service]` section: ```bash [Service] Environment="OLLAMA_MODELS=/path/to/ollama/models" ``` 3. Save and exit the editor. ### 5. Reload and Restart the Ollama Service Apply the changes by reloading the system configuration and restarting the service. ```bash sudo systemctl daemon-reload sudo systemctl restart ollama ``` ### 6. Verify the Configuration Ensure the service is running without errors: ```bash sudo systemctl status ollama ``` Test the connectivity: ```bash curl http://localhost:11434/api/version ``` If the service responds with the version details, the configuration is successful.
GiteaMirror commented 2026-04-12 13:06:38 -05:00
Author
Owner
Copy Link

@donymorph commented on GitHub (Mar 18, 2025):

I don't know. Environment="OLLAMA_MODELS=/path/to/ollama/models" didnt work for me. Instead I used symbolic link
ln -s model/stored/path ollama/recognized/path
example
sudo ln -s /media/hdd/models /home/officepc/.ollama

<!-- gh-comment-id:2733534151 --> @donymorph commented on GitHub (Mar 18, 2025): I don't know. Environment="OLLAMA_MODELS=/path/to/ollama/models" didnt work for me. Instead I used symbolic link `ln -s model/stored/path ollama/recognized/path` example `sudo ln -s /media/hdd/models /home/officepc/.ollama`
GiteaMirror commented 2026-04-12 13:06:39 -05:00
Author
Owner
Copy Link

@abubakr380 commented on GitHub (Aug 26, 2025):

My problem was with the permissions of the parent models folder. It was initially created by ollama user, and I later changed the user in systemd file. So now, that new user doesn't have write permission (ollama executes mkdir to make sure the parent directory exists).
To check permissions: sudo ls -la /usr/share/ollama/.ollama

Solution: Change directory owner to new user sudo chown -R new_user /usr/share/ollama/ (replace new_user)

<!-- gh-comment-id:3224059836 --> @abubakr380 commented on GitHub (Aug 26, 2025): My problem was with the permissions of the parent models folder. It was initially created by `ollama` user, and I later changed the user in systemd file. So now, that new user doesn't have write permission (ollama executes mkdir to make sure the parent directory exists). To check permissions: `sudo ls -la /usr/share/ollama/.ollama` Solution: Change directory owner to new user `sudo chown -R new_user /usr/share/ollama/` (replace new_user)
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 bug
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#2783
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?