github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-16 13:19:40 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #7284] Is default install location configurable #4628

New Issue
Closed
opened 2026-04-12 15:32:19 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-12 15:32:19 -05:00
Owner
Copy Link

Originally created by @wgong on GitHub (Oct 21, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/7284

I run Ollama on Ubuntu, where its default home is /usr/share/ollama.
Recently, I ran out of space on that partition after experimenting with quite a few models.
I then modified the install.sh script to install Ollama to /opt/ollama.
This worked when starting Ollama on the terminal by running ollama serve; however, when I tried running it as a systemctl service,
it failed. Checking the log via sudo journalctl -u ollama -n 50 --no-pager
showed an error message like "no permission to write id file in /usr/share folder".
I'm curious if /usr/share is hard-coded somewhere.

After creating a symlink for /usr/share/ollama to /opt/ollama and revising /etc/systemd/system/ollama.service to run as root, I was finally able to start the systemctl service.
Although it worked for me, I suspect it's a hack. Could you please advise on the correct way to install Ollama to a custom location and run it as a service (not as root)?

Claude Pro helped me with the above troubleshooting and workaround.
It mentioned that the following environment variables are customizable. Is this correct?

OLLAMA_HOST: Specifies the host and port Ollama listens on (default: 127.0.0.1:11434).
OLLAMA_MODELS: We've discussed this; it sets the directory for model storage.
OLLAMA_HOME: Sets the home directory for Ollama (default: ~/.ollama).
OLLAMA_KEEP_ALIVE: Sets the duration to keep models loaded in memory (default: 5m).
OLLAMA_TIMEOUT: Sets the timeout for loading models (default: 5m).
OLLAMA_ORIGINS: Comma-separated list of allowed origins for CORS.
OLLAMA_NOPRUNE: If set, prevents automatic removal of unused layers.
OLLAMA_DEBUGLOG: Enables debug logging when set to 'on'.
OLLAMA_VERBOSE: Enables verbose logging when set to 'on'.
OLLAMA_GPU_LAYERS: Number of layers to offload to GPU (depends on available VRAM).
OLLAMA_THREADS: Sets the number of threads to use (default is number of physical CPU cores).
OLLAMA_SKIP_UPDATE_CHECK: Skips the update check on startup if set.
OLLAMA_INSECURE: Allows insecure connections when pulling models if set.
OLLAMA_LOCKFILE: Specifies a custom path for the lockfile.
OLLAMA_LOGFILE: Specifies a custom path for the logfile.

Thanks

Originally created by @wgong on GitHub (Oct 21, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/7284 I run Ollama on Ubuntu, where its default home is `/usr/share/ollama`. Recently, I ran out of space on that partition after experimenting with quite a few models. I then modified the `install.sh` script to install Ollama to `/opt/ollama`. This worked when starting Ollama on the terminal by running `ollama serve`; however, when I tried running it as a systemctl service, it failed. Checking the log via `sudo journalctl -u ollama -n 50 --no-pager` showed an error message like "no permission to write id file in /usr/share folder". I'm curious if `/usr/share` is hard-coded somewhere. After creating a symlink for `/usr/share/ollama` to `/opt/ollama` and revising `/etc/systemd/system/ollama.service` to run as root, I was finally able to start the systemctl service. Although it worked for me, I suspect it's a hack. Could you please advise on the correct way to install Ollama to a custom location and run it as a service (not as root)? Claude Pro helped me with the above troubleshooting and workaround. It mentioned that the following environment variables are customizable. Is this correct? ``` OLLAMA_HOST: Specifies the host and port Ollama listens on (default: 127.0.0.1:11434). OLLAMA_MODELS: We've discussed this; it sets the directory for model storage. OLLAMA_HOME: Sets the home directory for Ollama (default: ~/.ollama). OLLAMA_KEEP_ALIVE: Sets the duration to keep models loaded in memory (default: 5m). OLLAMA_TIMEOUT: Sets the timeout for loading models (default: 5m). OLLAMA_ORIGINS: Comma-separated list of allowed origins for CORS. OLLAMA_NOPRUNE: If set, prevents automatic removal of unused layers. OLLAMA_DEBUGLOG: Enables debug logging when set to 'on'. OLLAMA_VERBOSE: Enables verbose logging when set to 'on'. OLLAMA_GPU_LAYERS: Number of layers to offload to GPU (depends on available VRAM). OLLAMA_THREADS: Sets the number of threads to use (default is number of physical CPU cores). OLLAMA_SKIP_UPDATE_CHECK: Skips the update check on startup if set. OLLAMA_INSECURE: Allows insecure connections when pulling models if set. OLLAMA_LOCKFILE: Specifies a custom path for the lockfile. OLLAMA_LOGFILE: Specifies a custom path for the logfile. ``` Thanks
GiteaMirror added the question label 2026-04-12 15:32:19 -05:00
GiteaMirror commented 2026-04-12 15:32:20 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 21, 2024):

Set OLLAMA_MODELS=/path/to/a/partition/with/space in the service override config. chown ollama /path/to/a/partition/with/space. ollama will use /usr/share/ollama to store key and history, all of the space consuming model files will be in /path/to/a/partition/with/space.

<!-- gh-comment-id:2427382121 --> @rick-github commented on GitHub (Oct 21, 2024): Set `OLLAMA_MODELS=/path/to/a/partition/with/space` in the service override config. `chown ollama /path/to/a/partition/with/space`. ollama will use `/usr/share/ollama` to store key and history, all of the space consuming model files will be in `/path/to/a/partition/with/space`.
GiteaMirror commented 2026-04-12 15:32:21 -05:00
Author
Owner
Copy Link

@wgong commented on GitHub (Oct 22, 2024):

Set OLLAMA_MODELS=/path/to/a/partition/with/space in the service override config. chown ollama /path/to/a/partition/with/space. ollama will use /usr/share/ollama to store key and history, all of the space consuming model files will be in /path/to/a/partition/with/space.

Thank you, will try it out

<!-- gh-comment-id:2428043677 --> @wgong commented on GitHub (Oct 22, 2024): > Set `OLLAMA_MODELS=/path/to/a/partition/with/space` in the service override config. `chown ollama /path/to/a/partition/with/space`. ollama will use `/usr/share/ollama` to store key and history, all of the space consuming model files will be in `/path/to/a/partition/with/space`. Thank you, will try it out
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 question
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#4628
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?