github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 00:51:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #12507] Make ollama pull save download progress #70363

New Issue
Closed
opened 2026-05-04 21:16:49 -05:00 by GiteaMirror · 7 comments
GiteaMirror commented 2026-05-04 21:16:49 -05:00
Owner
Copy Link

Originally created by @TracerDS on GitHub (Oct 5, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/12507

As in the title. If you download a large model it will take a long time for it to download.
Saving the download progress would be awesome if your network isnt the fastest.

Originally created by @TracerDS on GitHub (Oct 5, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/12507 As in the title. If you download a large model it will take a long time for it to download. Saving the download progress would be awesome if your network isnt the fastest.
GiteaMirror added the feature request label 2026-05-04 21:16:49 -05:00
GiteaMirror commented 2026-05-04 21:16:53 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 5, 2025):

If a download is interrupted (by you or the connection failing) ollama will keep the downloaded pieces until the server is restarted. You can resume the download by re-running the pull command. If you want to be able to restart the server and continue to keep downloaded chunks, set OLLAMA_NOPRUNE=1 in the server environment.

<!-- gh-comment-id:3369113182 --> @rick-github commented on GitHub (Oct 5, 2025): If a download is interrupted (by you or the connection failing) ollama will keep the downloaded pieces until the server is restarted. You can resume the download by re-running the pull command. If you want to be able to restart the server and continue to keep downloaded chunks, set `OLLAMA_NOPRUNE=1` in the server environment.
GiteaMirror commented 2026-05-04 21:16:53 -05:00
Author
Owner
Copy Link

@TracerDS commented on GitHub (Oct 5, 2025):

Frankly I dont think thats enough.
Sure, thats an option but its very unintuitive and the chunks arent saved 1:1 (i.e. you download 70% and 70% remains on your machine) but rather with some "delay".
It would be even better if there was an option that shows you how much % of the model you have downloaded (when not downloading atm) but its a small qol feature.

By "until the server is restarted" you mean the ollama pull ... is executed again?

<!-- gh-comment-id:3369261770 --> @TracerDS commented on GitHub (Oct 5, 2025): Frankly I dont think thats enough. Sure, thats an option but its very unintuitive and the chunks arent saved 1:1 (i.e. you download 70% and 70% remains on your machine) but rather with some "delay". It would be even better if there was an option that shows you how much % of the model you have downloaded (when not downloading atm) but its a small qol feature. By "until the server is restarted" you mean the `ollama pull ...` is executed again?
GiteaMirror commented 2026-05-04 21:16:54 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 5, 2025):

Sure, thats an option but its very unintuitive and the chunks arent saved 1:1 (i.e. you download 70% and 70% remains on your machine) but rather with some "delay".

The chunks that are completely downloaded remain on the machine. If 70% of the chunks are downloaded and the chunk that marks 71% is starting download when the downloaded is interrupted, that 1% will be re-downloaded (along with the other 29%) the next time ollama pull is run. The 70% downloaded chunks remain until the server is restarted or the pull is finished.

By "until the server is restarted" you mean the ollama pull ... is executed again?

Restarting means stopping and starting the ollama server. Rebooting the computer is an example. Clicking on the ollama systray entry and then clicking on "Restart to update", or on "Quit ollama" and then running the ollama app, also counts as restarting the server.

<!-- gh-comment-id:3369271321 --> @rick-github commented on GitHub (Oct 5, 2025): > Sure, thats an option but its very unintuitive and the chunks arent saved 1:1 (i.e. you download 70% and 70% remains on your machine) but rather with some "delay". The chunks that are completely downloaded remain on the machine. If 70% of the chunks are downloaded and the chunk that marks 71% is starting download when the downloaded is interrupted, that 1% will be re-downloaded (along with the other 29%) the next time `ollama pull` is run. The 70% downloaded chunks remain until the server is restarted or the pull is finished. > By "until the server is restarted" you mean the ollama pull ... is executed again? Restarting means stopping and starting the ollama server. Rebooting the computer is an example. Clicking on the ollama systray entry and then clicking on "Restart to update", or on "Quit ollama" and then running the ollama app, also counts as restarting the server.
GiteaMirror commented 2026-05-04 21:16:55 -05:00
Author
Owner
Copy Link

@TracerDS commented on GitHub (Oct 6, 2025):

If a user unintentionally starts to pull a model and changes their mind mid download and wants to stop the model being pulled with keyboard interrupt, they should also have an option to clear the incomplete downloads to save the disc space or It'll be an unnecessary bloat.

Absolutely, but for those who do want to download a model this isnt an ideal option really

<!-- gh-comment-id:3370364852 --> @TracerDS commented on GitHub (Oct 6, 2025): > If a user unintentionally starts to pull a model and changes their mind mid download and wants to stop the model being pulled with keyboard interrupt, they should also have an option to `clear` the incomplete downloads to save the disc space or It'll be an unnecessary bloat. Absolutely, but for those who do want to download a model this isnt an ideal option really
GiteaMirror commented 2026-05-04 21:16:56 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 6, 2025):

this would be a soft-delete the current implementation is a hard-delete and re-restart every-time when model download fails or stalls

The current implementation is not hard-delete and restart when the download fails or stalls. If the download is interrupted, the currently downloaded portion of the model is preserved until the pull is resumed or the ollama server is restarted.

If you would like to do download management, see https://github.com/ollama/ollama/issues/12262#issuecomment-3284822994 for a script that allows downloading a model outside of the model directory, and supports download restarts in the case of interruption.

<!-- gh-comment-id:3370964027 --> @rick-github commented on GitHub (Oct 6, 2025): > this would be a soft-delete the current implementation is a hard-delete and re-restart every-time when model download fails or stalls The current implementation is not hard-delete and restart when the download fails or stalls. If the download is interrupted, the currently downloaded portion of the model is preserved until the pull is resumed or the ollama server is restarted. If you would like to do download management, see https://github.com/ollama/ollama/issues/12262#issuecomment-3284822994 for a script that allows downloading a model outside of the model directory, and supports download restarts in the case of interruption.
GiteaMirror commented 2026-05-04 21:16:56 -05:00
Author
Owner
Copy Link

@pdevine commented on GitHub (Oct 9, 2025):

@yashwanth2706 as @rick-github mentioned, you can just restart the server (or your computer), and it will clear up any leftover pulls. When I wrote the feature I wanted it to be simple without having to have yet another command and more flags.

I'm going to go ahead and close the issue as answered as the described feature already exists.

<!-- gh-comment-id:3383764055 --> @pdevine commented on GitHub (Oct 9, 2025): @yashwanth2706 as @rick-github mentioned, you can just restart the server (or your computer), and it will clear up any leftover pulls. When I wrote the feature I wanted it to be simple without having to have yet another command and more flags. I'm going to go ahead and close the issue as answered as the described feature already exists.
GiteaMirror commented 2026-05-04 21:16:56 -05:00
Author
Owner
Copy Link

@peg-leg commented on GitHub (Apr 16, 2026):

Am I missing something? Does Ollama run not follow the same process as pull?

Image

It starts from scratch every time on my PC...

<!-- gh-comment-id:4258529247 --> @peg-leg commented on GitHub (Apr 16, 2026): Am I missing something? Does Ollama run not follow the same process as pull? <img width="2390" height="890" alt="Image" src="https://github.com/user-attachments/assets/64346ae8-93ef-4233-b60b-58c41627eeec" /> It starts from scratch every time on my PC...
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 feature request
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#70363
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?