Hacktoberfest 2026：维护者为十月标记出来的 issue，仍然开放、适合新手。 浏览 Hacktoberfest issue

[Bug] RaftEngineReadIndexTest fails now and then with UnfinishedStubbingException: the alive-peers refresher can call the mocked raft node while a test stubs it

已关闭
  • 哪里不对？
#3,384 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

已经有一个关联 PR 被合并了。

  • #3385 来自 @Mihir-Rabari —— 已合并

评估

难度
4/5
预计耗时
3-5 天
新手友好度
35/100
Issue 类型
缺陷
描述清晰度
描述清楚
活跃度
停滞
技术栈
java
领域
databases, testing

调研方向

Read RaftEngineReadIndexTest.java around answerShutdown and testStoppedReadServiceClosureThatRunsFirstFailsAsNotLeader, then inspect RaftEngine.startAlivePeersRefresher and its listAlivePeers call. Reproduce with the stated Maven command and determine whether the refresher can call the mock during stubbing. The linked pull request #3385 is already open; check its status and whether it resolves the intermittent failure.

由索引模型根据 Issue 内容生成。

描述

Bug Type (问题类型)

others (please edit later)

Before submit
  • I have confirmed and searched that there are no similar problems in the historical issue and documents
Environment (环境信息)
  • Server Version: PD tests, hg-pd-test pd-core-test profile (PDCoreSuiteTest). Seen on the branch of #3376 at 171760d25, which does not change RaftEngineReadIndexTest or the alive-peers refresher of RaftEngine; both are as on master 86c05ef9c (the test from #3368, the refresher from #3185)
  • Backend: none (unit suite)
  • OS: Linux, 8 cores, JDK 17
  • Data Size: N/A
Expected & Actual behavior (期望与实际表现)

RaftEngineReadIndexTest.testStoppedReadServiceClosureThatRunsFirstFailsAsNotLeader should pass every run. It failed once in three runs of the same commit on 2026-10-20, with Mockito's unfinished-stubbing error while the test stubs readIndex:

[ERROR] org.apache.hugegraph.pd.raft.RaftEngineReadIndexTest.testStoppedReadServiceClosureThatRunsFirstFailsAsNotLeader -- Time elapsed: 0.012 s <<< ERROR!
org.mockito.exceptions.misusing.UnfinishedStubbingException:
Unfinished stubbing detected here:
-> at org.apache.hugegraph.pd.raft.RaftEngineReadIndexTest.answerShutdown(RaftEngineReadIndexTest.java:377)
	at org.apache.hugegraph.pd.raft.RaftEngineReadIndexTest.answerShutdown(RaftEngineReadIndexTest.java:382)
	at org.apache.hugegraph.pd.raft.RaftEngineReadIndexTest.testStoppedReadServiceClosureThatRunsFirstFailsAsNotLeader(RaftEngineReadIndexTest.java:367)

The runs, all mvn -B -ntp test -pl hugegraph-pd/hg-pd-test -am -P pd-core-test on 171760d25:

Host pd-core (run/failures/errors/skipped)
8 cores, JDK 17 227/0/1/2 (this error)
same host, again 227/0/0/2
12 cores, JDK 17 227/0/0/2

Likely cause, not proven: the test swaps a Mockito mock into the RaftEngine singleton (Whitebox.setInternalState(..., "raftNode", mockNode)) and stubs it, while the engine's pd-raft-alive-peers refresher thread (startAlivePeersRefresher, every ALIVE_PEERS_REFRESH_MS) calls node.listAlivePeers() on whatever node the singleton holds whenever its state machine reports leader. An earlier suite test can leave the singleton started and leading. A call on the mock from that thread while doAnswer(...).when(mockNode).readIndex(...) is in progress is the case Mockito reports as unfinished stubbing.

A fix could stop the refresher for these tests (or give the refresher a node snapshot taken under the same lock the tests use), or stub listAlivePeers on the mock before it is installed.

Vertex/Edge example (问题点 / 边数据举例)

N/A

Schema [VertexLabel, EdgeLabel, IndexLabel] (元数据结构)

N/A

主要语言
Java
星标
3.2k
派生
640
平均合并
1 天 12 小时
30 天内合并 PR
98

环境准备

从这里开始

  1. 先读完整个 Issue，再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库，在一个分支上完成修改。
  4. 提交 Pull Request，并在描述里引用这个 Issue 编号。

apache/hugegraph 的其他 Issue

查看 apache/hugegraph 的全部 Issue

相似的 Issue

更多 Java Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。