github-starred/ollama-ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-03-11 20:23:55 -05:00
Code Issues 2.2k Packages Projects Releases 100 Wiki Activity

[PR #12616] llm: Optimize memory calculation on ZFS, Add env var control for memory check and log level #12624

New Issue
Open
opened 2025-11-12 16:41:12 -06:00 by GiteaMirror · 0 comments
GiteaMirror commented 2025-11-12 16:41:12 -06:00
Owner
Copy Link

📋 Pull Request Information

Original PR: https://github.com/ollama/ollama/pull/12616
Author: @arunskurian
Created: 10/14/2025
Status: 🔄 Open

Base: main ← Head: zfs-memcheck

📝 Commits (6)

  • 2164189 Projects/2025.4.0.archive (#12)
  • 47b7b47 DLPXECO-12676 Build error in security.go loadfunction
  • 046e33c Upgrading ollama version (#13)
  • b89f481 Merge branch 'ollama:main' into main
  • a10d926 Merge branch 'ollama:main' into main
  • 1324148 Consolidate memory check snippet. Remove CPU build files (#14)

📊 Changes

6 files changed (+117 additions, -9 deletions)

View changed files

📝 cmd/cmd.go (+1 -0)
📝 docs/faq.md (+20 -0)
📝 envconfig/config.go (+9 -0)
📝 llm/memory.go (+54 -0)
📝 llm/server.go (+25 -8)
📝 server/routes.go (+8 -1)

📄 Description

Context

This PR addresses the following 2 issues

1. Free Memory Calculation when running on ZFS + Env variable to bypass memory checks

(https://github.com/ollama/ollama/issues/5700)

For the Free Memory Calculation when running on ZFS, two changes have been introduced.

  1. By setting the OLLAMA_SKIP_MEMORY_CHECK environment variable to 1, this version can override Ollama's available system memory check. *Only for advanced use cases and must be used with caution
    If not set, memory checks will progress BAU.

  2. On ZFS machines, we now calculate the amount of memory that can be evicted from the Arc Cache, which allows us to proceed to loading the models. ZFS resizes the cache when we claim the memory thats captured by Arch Cache. However, there are guardrails on how much memory we can evict for use. This amount is calculated based on info read from zfs arcstats file.
    On non ZFS systems ( or when the zfs config cannot be read), the memory calculation will proceed as BAU.

2. Logging based on GIN_MODE

https://github.com/ollama/ollama/issues/8339

🔄 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/12616 **Author:** [@arunskurian](https://github.com/arunskurian) **Created:** 10/14/2025 **Status:** 🔄 Open **Base:** `main` ← **Head:** `zfs-memcheck` --- ### 📝 Commits (6) - [`2164189`](https://github.com/ollama/ollama/commit/2164189200d5be638b4eeb2d9578ef7fa5645d46) Projects/2025.4.0.archive (#12) - [`47b7b47`](https://github.com/ollama/ollama/commit/47b7b473b32d7418ec4cc157db28a59d849475dd) DLPXECO-12676 Build error in security.go loadfunction - [`046e33c`](https://github.com/ollama/ollama/commit/046e33cea1b38f38fc05074721191f85da4e91c3) Upgrading ollama version (#13) - [`b89f481`](https://github.com/ollama/ollama/commit/b89f4815578fc865c6a45cb3e4287157beadfa47) Merge branch 'ollama:main' into main - [`a10d926`](https://github.com/ollama/ollama/commit/a10d926b5fd52d93f25c03f9553833a06e5b226a) Merge branch 'ollama:main' into main - [`1324148`](https://github.com/ollama/ollama/commit/13241488c21a6bd53905a84bcb6821ce9b006034) Consolidate memory check snippet. Remove CPU build files (#14) ### 📊 Changes **6 files changed** (+117 additions, -9 deletions) <details> <summary>View changed files</summary> 📝 `cmd/cmd.go` (+1 -0) 📝 `docs/faq.md` (+20 -0) 📝 `envconfig/config.go` (+9 -0) 📝 `llm/memory.go` (+54 -0) 📝 `llm/server.go` (+25 -8) 📝 `server/routes.go` (+8 -1) </details> ### 📄 Description ## Context This PR addresses the following 2 issues ### 1. Free Memory Calculation when running on ZFS + Env variable to bypass memory checks (https://github.com/ollama/ollama/issues/5700) For the Free Memory Calculation when running on ZFS, two changes have been introduced. 1. By setting the OLLAMA_SKIP_MEMORY_CHECK environment variable to 1, this version can override Ollama's available system memory check. *Only for advanced use cases and must be used with caution If not set, memory checks will progress BAU. 2. On ZFS machines, we now calculate the amount of memory that can be evicted from the Arc Cache, which allows us to proceed to loading the models. ZFS resizes the cache when we claim the memory thats captured by Arch Cache. However, there are guardrails on how much memory we can evict for use. This amount is calculated based on info read from zfs arcstats file. On non ZFS systems ( or when the zfs config cannot be read), the memory calculation will proceed as BAU. ### 2. Logging based on GIN_MODE https://github.com/ollama/ollama/issues/8339 --- <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 2025-11-12 16:41:12 -06:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
client2
cloud
compatibility
context-length
create
docker
documentation
embeddings
engine
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
install
integration
intel
js
linux
macos
memory
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
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-ollama#12624
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?