alibaba/GraphScope

[GLE] Explicitly build the training client using the GLE server handle

Open

#2,738 opened on May 25, 2023

View on GitHub
 (1 comment) (0 reactions) (1 assignee)HTML (301 forks)batch import
component:glegood first issue

Repository metrics

Stars
 (2,401 stars)
PR merge metrics
 (Avg merge 1m) (7 merged PRs in 30d)

Description

A graph object is returned after the GLE servers are launched, which is further used to build the client that the training worker uses to communicate with the GLE servers. While in distributed training（multiple training workers), we need to first get the handle of the graph (including the schema info, service IP and port), and use these information to build client in the training side. This issue is to add APIs and implementations to ease the process of explicitly building the training client using the GLE server handle.

Contributor guide

Research direction
Review the existing GLE server handle object to understand its structure. Look at the current client building process for single worker training. Then design an API to extract handle information (schema, IP, port) and expose it. Check any existing similar APIs in the codebase. Consider the comments in the issue for guidance.
Tech stack
pythoncpp
Domain
backendapimachine learning
Issue type
Feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
3-5 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
Blocked
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
Clear
Prerequisites
Distributed systems understandingPython programmingFamiliarity with GraphScope GLE
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
20