lablup / lablup/backend.ai

Migrate resource allocation map from agent to manager

Open
#3,515 0 comments 0 reactions 1 assignee Claimed by @fregataa View on GitHub
Dominant language
Python
Stars
670
Forks
183
Avg merge
17h 7m
Merged PRs (30d)
358

Description

## Motivation  

- `InsufficientResource` errors occur in the CREATING phase of session creation, while creating containers in agents. This error may result from:
- Resource state mismatches between managers and agents ([BA-586](https://lablup.atlassian.net/browse/BA-586))
- **Resource fragmentation in agent**.
Since managers don't have visibility into agents' resource allocation maps, they can only schedule sessions by comparing the required resource amount with the total sum of remaining resources in agents. This leads to scheduling sessions to inappropriate agents that have fragmented resources

## Objective  

- Let scheduler consider allocation map and avoid scheduling sessions to resource fragmented agents

## Expected Sub Issue

- Migrate resource allocation map from agent to manager
- Update schedulers of manager to consider the allocation map
- Update all accelerator plugins

JIRA Issue: BA-588

[BA-586]: https://lablup.atlassian.net/browse/BA-586?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.