mirror of
https://github.com/deepseek-ai/3FS.git
synced 2026-11-05 00:47:47 +00:00
[PR #350] [MERGED] update default value of max_idle_threads to keep consistent with libfuse #383
Labels
No labels
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
deepseek-ai/3FS#383
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
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?
📋 Pull Request Information
Original PR: https://github.com/deepseek-ai/3FS/pull/350
Author: @nautaa
Created: 10/14/2025
Status: ✅ Merged
Merged: 10/14/2025
Merged by: @SF-Zhou
Base:
main← Head:update_max_idle_threads📝 Commits (1)
a8beb55update default value of max_idle_threads to keep consistent with libfuse📊 Changes
2 files changed (+2 additions, -2 deletions)
View changed files
📝
configs/hf3fs_fuse_main.toml(+1 -1)📝
src/fuse/FuseConfig.h(+1 -1)📄 Description
During our testing(fio sequential read), we observed significant thread creation overhead, which ultimately traced to the creation of fuse_worker threads within libfuse. In libfuse, the default value for max_idle_threads is -1, and as the libfuse comments, it is recommended to set max_idle_threads to -1. max_idle_threads means the maximum number of idle worker threads. After exceeding max_idle_threads active threads, libfuse destroys redundant idle threads. This may result in frequent thread creation and destruction.
However, in 3FS, this value defaults to 10. In our test environment, setting max_idle_threads = 10 resulted in considerable overhead from thread creation and destruction. Therefore, I believe aligning the default value of max_idle_threads in 3FS with that of libfuse would be a more reasonable default choice.
github.com/libfuse/libfuse@564d79e4d3/lib/fuse_loop_mt.c (L185-L190)🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.