github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 16:40:08 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #1687] Old Models disappear after Ollama Update (0.1.17) #62986

New Issue
Closed
opened 2026-05-03 11:07:49 -05:00 by GiteaMirror · 8 comments
GiteaMirror commented 2026-05-03 11:07:49 -05:00
Owner
Copy Link

Originally created by @sthufnagl on GitHub (Dec 23, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/1687

Hi,

Environment:
my environment ist WSL on Win11.

Update Command:
curl https://ollama.ai/install.sh | sh

Situation:
After an Update to Ollama 0.1.17 all my old Models (202GB) are not visible anymore and when I try to start an old one the Model is downloaded once again. Physically the Model Files are available but not listed (ollama list) or used.
I want to avoid to download the old models once again.

Reason for Update:
I wanted to use PHI but after downloading I got some error messages ==> Update of ollama was necessary

Question:

  • Can I restore my old Models?
  • Is there an entry in a config file?

Thx in advance

Originally created by @sthufnagl on GitHub (Dec 23, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/1687 Hi, **Environment:** my environment ist WSL on Win11. **Update Command:** curl https://ollama.ai/install.sh | sh **Situation:** After an Update to Ollama 0.1.17 all my old Models (202GB) are not visible anymore and when I try to start an old one the Model is downloaded once again. Physically the Model Files are available but not listed (ollama list) or used. I want to avoid to download the old models once again. **Reason for Update:** I wanted to use PHI but after downloading I got some error messages ==> Update of ollama was necessary **Question:** * Can I restore my old Models? * Is there an entry in a config file? Thx in advance
GiteaMirror added the bug label 2026-05-03 11:07:49 -05:00
GiteaMirror commented 2026-05-03 11:07:54 -05:00
Author
Owner
Copy Link

@coffeecodechem commented on GitHub (Dec 23, 2023):

Physically the model is in GGUF right? Maybe try this guide: https://github.com/jmorganca/ollama?tab=readme-ov-file#import-from-gguf

Then for each model you have, create a Modelfile https://github.com/jmorganca/ollama/blob/main/docs/modelfile.md

You will find the recommended template for each model in https://ollama.ai/library. Look for the model, click Tags.

<!-- gh-comment-id:1868284366 --> @coffeecodechem commented on GitHub (Dec 23, 2023): Physically the model is in GGUF right? Maybe try this guide: https://github.com/jmorganca/ollama?tab=readme-ov-file#import-from-gguf Then for each model you have, create a Modelfile https://github.com/jmorganca/ollama/blob/main/docs/modelfile.md You will find the recommended template for each model in https://ollama.ai/library. Look for the model, click Tags.
GiteaMirror commented 2026-05-03 11:07:56 -05:00
Author
Owner
Copy Link

@BruceMacD commented on GitHub (Dec 24, 2023):

There seems like there could be an on-going bug here, this sounds similar to #1493. In that case they were able to get Ollama to pick-up the models by running "list" a few times.

For troubleshooting reasons which directory are you seeing the model files in that Ollama is not picking up?

<!-- gh-comment-id:1868527911 --> @BruceMacD commented on GitHub (Dec 24, 2023): There seems like there could be an on-going bug here, this sounds similar to #1493. In that case they were able to get Ollama to pick-up the models by running "list" a few times. For troubleshooting reasons which directory are you seeing the model files in that Ollama is not picking up?
GiteaMirror commented 2026-05-03 11:07:58 -05:00
Author
Owner
Copy Link

@sthufnagl commented on GitHub (Dec 25, 2023):

Hi,
thanks for your helping hand.
:-)
Because my old PC had watercooling problems...I installed ollama on a different PC.
That is the reason that I cann't test your recommendations...
:-(

But I recognize that with 0.1.17 the models are stored on a different location.

Ollama 0.1.16 use: ~/.ollama/models
Ollama 0.1.17 use: /usr/share/ollama/.ollama/models

Maybe this is the reason?

BTW, I made a backup on an USB Stick before I did an Ollama Upgrade.
Can I simple copy the model files to /usr/share/ollama/.ollama/models ?

Thx

<!-- gh-comment-id:1869034828 --> @sthufnagl commented on GitHub (Dec 25, 2023): Hi, thanks for your helping hand. :-) Because my old PC had watercooling problems...I installed ollama on a different PC. That is the reason that I cann't test your recommendations... :-( But I recognize that with 0.1.17 the **models are stored on a different location**. **Ollama 0.1.16 use:** ~/.ollama/models **Ollama 0.1.17 use:** /usr/share/ollama/.ollama/models Maybe this is the reason? BTW, I made a backup on an USB Stick before I did an Ollama Upgrade. Can I simple copy the model files to /usr/share/ollama/.ollama/models ? Thx
GiteaMirror commented 2026-05-03 11:08:01 -05:00
Author
Owner
Copy Link

@sthufnagl commented on GitHub (Dec 26, 2023):

Hi,
I'm now one step further...and find a workaround.

  • It seems that Ollama 0.1.17 install a system service "ollama.service"
  • This service expect the model files at /usr/share/ollama/.ollama/models
  • When I stop the service with "systemctl stop ollama.service" and start Ollama with "ollama serve &" Ollama expect the Model Files at "~/.ollama/models"
  • After restoring the Model Files from USB Stick to "~/.ollama/models" everything works!!!
  • So I will disable the Ollama system service

Thx

<!-- gh-comment-id:1869497028 --> @sthufnagl commented on GitHub (Dec 26, 2023): Hi, I'm now one step further...and find a workaround. * It seems that Ollama 0.1.17 install a system service "ollama.service" * This service expect the model files at /usr/share/ollama/.ollama/models * When I stop the service with "systemctl stop ollama.service" and start Ollama with "ollama serve &" Ollama expect the Model Files at "~/.ollama/models" * After restoring the Model Files from USB Stick to "~/.ollama/models" everything works!!! * So I will disable the Ollama system service Thx
GiteaMirror commented 2026-05-03 11:08:03 -05:00
Author
Owner
Copy Link

@sthufnagl commented on GitHub (Dec 26, 2023):

Please see last comment...

<!-- gh-comment-id:1869497322 --> @sthufnagl commented on GitHub (Dec 26, 2023): Please see last comment...
GiteaMirror commented 2026-05-03 11:08:08 -05:00
Author
Owner
Copy Link

@BruceMacD commented on GitHub (Dec 27, 2023):

Thanks for following up 🙏

<!-- gh-comment-id:1870494834 --> @BruceMacD commented on GitHub (Dec 27, 2023): Thanks for following up 🙏
GiteaMirror commented 2026-05-03 11:08:12 -05:00
Author
Owner
Copy Link

@davidbuzz commented on GitHub (Jul 19, 2024):

I consider this a bug that 'ollama serve' and the ollama service look in different folder/s for their models. At the very least 'ollama serve' should be able to look at both folder/s.

<!-- gh-comment-id:2238528569 --> @davidbuzz commented on GitHub (Jul 19, 2024): I consider this a bug that 'ollama serve' and the ollama service look in different folder/s for their models. At the very least 'ollama serve' should be able to look at both folder/s.
GiteaMirror commented 2026-05-03 11:08:20 -05:00
Author
Owner
Copy Link

@little-engineer-2025 commented on GitHub (Aug 19, 2025):

I faced similar situation; I started and downloaded a model by running ollama serve with my user, and thanks to this thread (thanks @sthufnagl ) I found where was downloaded the model; In my scenario I found that only moving the ~/.ollama/{blobs,manifests} to /usr/share/ollama/{blobs,manifests} respectively, and changing the owner to ollama:ollama, I could run as a systemd service and see the model was available by ollama list and use it by just ollama run mymodel. I was using Arch linux, maybe that made some path different; Sharing if it helps to other users.

<!-- gh-comment-id:3201645676 --> @little-engineer-2025 commented on GitHub (Aug 19, 2025): I faced similar situation; I started and downloaded a model by running `ollama serve` with my user, and thanks to this thread (thanks @sthufnagl ) I found where was downloaded the model; In my scenario I found that only moving the ~/.ollama/{blobs,manifests} to /usr/share/ollama/{blobs,manifests} respectively, and changing the owner to `ollama:ollama`, I could run as a systemd service and see the model was available by `ollama list` and use it by just `ollama run mymodel`. I was using Arch linux, maybe that made some path different; Sharing if it helps to other users.
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#62986
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?