Lightning-AI / Lightning-AI/LitServe

Allow resource release at the end of life of the LitAPI object

Open
#420 3 comments 1 reaction 0 assignees View on GitHub

Nobody has claimed this yet.

enhancement won't fix
Dominant language
Python
Stars
3.9k
Forks
304
Avg merge
3d 13h
Merged PRs (30d)
6

Description


🚀 Feature

We have a setup method, which allows one to initialize for each worker the resources needed for inference.

I want the exact reverse: When the server closes, to release the resource (via an abstract teardown method or similar)

Motivation

Some APIs need more than just temporary state (DB connection, prompt files, etc...). Not being able to release them forces one to use garbage collection to release the data, which is an unsafe way to proceed.

Pitch
sequenceDiagram
    participant MAIN as parent process
    create participant WORKER as worker process
    MAIN ->> WORKER: fork()
    WORKER ->> WORKER: setup
    note over MAIN, WORKER: server lifetime
    WORKER ->> WORKER: teardown
    destroy WORKER
    WORKER ->> MAIN: child process is dead
    MAIN ->> MAIN: program finalization

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by tracing the existing setup method through the worker lifecycle and locate the server-shutdown path. Define the teardown lifecycle and verify that it runs for each worker before the process exits, releasing resources such as database connections or prompt files.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
api, backend
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.