github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-15 20:54:54 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[PR #13341] server: add optional setting that prevents unloading currently loaded models to make room for a new model #24710

New Issue
Open
opened 2026-04-19 17:45:28 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-04-19 17:45:28 -05:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/13341
Author: @YetheSamartaka
Created: 12/5/2025
Status: 🔄 Open

Base: main ← Head: no-model-evict

📝 Commits (3)

  • 2515127 Add optional setting for preventing unloading current models for new requests
  • 7b202a9 Merge branch 'main' into no-model-evict
  • 5ab8acd Merge branch 'main' into no-model-evict

📊 Changes

5 files changed (+93 additions, -0 deletions)

View changed files

📝 cmd/cmd.go (+1 -0)
📝 docs/faq.mdx (+1 -0)
📝 envconfig/config.go (+3 -0)
📝 server/sched.go (+27 -0)
📝 server/sched_test.go (+61 -0)

📄 Description

Currently there is no way how to create a ollama docker container with pre-defined ollama models that will be there preloaded and stay there indefinitely mainly because if a request to the ollama instance would want to load different model, even tho the OLLAMA_MAX_LOADED_MODELS is set to a fixed number, Ollama will just unload one of those models to load new one, or in cases where this is not set but there is not enough vRAM, Ollama will unload some model to load the newly request one.

This is often unwanted. I introduced a new env variable OLLAMA_NO_MODEL_EVICT that when set to "1" will prevent those models from unloading in such events. In default settings this is disabled so it is completely optional. I believe this small change will make Ollama even more powerful.

I've tested it thoroughly and included tests for the current suite.

  • Added optional setting (OLLAMA_NO_MODEL_EVICT) for preventing unloading current models for new requests
  • Added tests for this feature

🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/ollama/ollama/pull/13341 **Author:** [@YetheSamartaka](https://github.com/YetheSamartaka) **Created:** 12/5/2025 **Status:** 🔄 Open **Base:** `main` ← **Head:** `no-model-evict` --- ### 📝 Commits (3) - [`2515127`](https://github.com/ollama/ollama/commit/251512721f3a6b8b968feda98cf3e053d3b8b1de) Add optional setting for preventing unloading current models for new requests - [`7b202a9`](https://github.com/ollama/ollama/commit/7b202a959d1b78a79f87dca2bba2f711c9963f17) Merge branch 'main' into no-model-evict - [`5ab8acd`](https://github.com/ollama/ollama/commit/5ab8acddc8aebb72573ca2251c02346b00aa7deb) Merge branch 'main' into no-model-evict ### 📊 Changes **5 files changed** (+93 additions, -0 deletions) <details> <summary>View changed files</summary> 📝 `cmd/cmd.go` (+1 -0) 📝 `docs/faq.mdx` (+1 -0) 📝 `envconfig/config.go` (+3 -0) 📝 `server/sched.go` (+27 -0) 📝 `server/sched_test.go` (+61 -0) </details> ### 📄 Description Currently there is no way how to create a ollama docker container with pre-defined ollama models that will be there preloaded and stay there indefinitely mainly because if a request to the ollama instance would want to load different model, even tho the OLLAMA_MAX_LOADED_MODELS is set to a fixed number, Ollama will just unload one of those models to load new one, or in cases where this is not set but there is not enough vRAM, Ollama will unload some model to load the newly request one. This is often unwanted. I introduced a new env variable OLLAMA_NO_MODEL_EVICT that when set to "1" will prevent those models from unloading in such events. In default settings this is disabled so it is completely optional. I believe this small change will make Ollama even more powerful. I've tested it thoroughly and included tests for the current suite. - Added optional setting (OLLAMA_NO_MODEL_EVICT) for preventing unloading current models for new requests - Added tests for this feature --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
GiteaMirror added the pull-request label 2026-04-19 17:45:28 -05:00
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 pull-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#24710
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?