pingcap / pingcap/tiflash

Validate TiFlash advertised endpoints before registering a new store with PD

Open
#10,977 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

component/storage type/bug
Dominant language
C++
Stars
1k
Forks
423
Avg merge
1d 15h
Merged PRs (30d)
24

Description

Bug Report

Problem

A TiFlash process can start with valid local listeners while advertising store, peer, and status endpoints that are intended for a different or future TiFlash instance.

In a reproduction, an orphan TiFlash registered first with:

[flash]
service_addr = "0.0.0.0:3931"

[flash.proxy]
addr = "0.0.0.0:20171"
advertise-addr = "127.0.0.1:20171"

[server]
engine-addr = "127.0.0.1:3931"
advertise-status-addr = "127.0.0.1:20293"

PD accepted the orphan as a TiFlash store because these addresses were not yet present. A later real TiFlash using the same intended endpoints failed during put store to PD with duplicated store address.

Impact

An orphan or misconfigured TiFlash can pre-claim a future store identity and block scale-out, replacement, or recovery. Stopping the process is not sufficient; the incorrect PD store metadata must be removed safely.

Expected behavior / possible improvement

Before registering with PD, TiFlash should validate that its advertised store, peer, and status endpoints route back to the same TiFlash instance, or emit a strong actionable warning when identity cannot be proven.

The validation should use an instance/store identity round trip where possible, rather than relying only on local interface matching, to remain compatible with container networking, DNS, Services, and load balancers.

Contributor guide

No contributing guide indexed for this repository

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

Begin with the reproduction configuration and the TiFlash put store to PD registration flow described in the issue. Trace how the advertised store, peer, and status endpoints are handled before registration. Done means preventing incorrect pre-claims through identity validation, or producing a strong actionable warning when identity cannot be proven.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
distributed-systems
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.