apache / apache/doris

[Row Binlog] Replica repair and balancing can break base/companion colocation and block writes

Open
#67,092 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Java
Stars
15.9k
Forks
3.9k
Avg merge
2d 23h
Merged PRs (30d)
522

Description

## Problem

[Row Binlog] Replica repair and balancing can break base/companion colocation and block writes

## Minimal reproduction path

Use a local cluster with at least five backends and a ROW Binlog table. Record the base/companion placement, increase replica count or move only base replicas until both sides are individually healthy but their colocated replica intersection is below write quorum, then run an INSERT or Stream Load. Verify that repair or scheduling restores paired placement, or reports a clear recoverable placement state instead of failing later with a replica-quorum error.

## Expected result

Replica repair and balancing preserve the required base/companion colocated quorum, or fail early with an actionable state.

## Source

This report is a sanitized tracking record for Jira DORIS-28061. Internal hosts, paths, artifacts, build identifiers, and private logs are intentionally omitted.

Contributor guide

Open the contributing guide

Research direction

Start with a local cluster of at least five backends and a ROW Binlog table, recording base and companion replica placement. Reproduce the repair or balancing scenario, then test INSERT or Stream Load; done means paired placement preserves write quorum or the system reports an actionable recoverable placement state.

Written by the indexing model from the issue text.

Assessment

Tech stack
java
Domain
databases, distributed-systems
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.