client-v2 retries an insert with the same query id after NoHttpResponseException, causing QUERY_WITH_SAME_ID_IS_ALREADY_RUNNING

未關閉
  • 哪裡不對？
#3,174 2 則留言 0 個 reaction 已指派 0 人 在 GitHub 檢視

維護者通常 1 天內回覆

還沒有人認領這個 Issue。

評估

難度
4/5
預估耗時
3-5 天
新手友好度
55/100
Issue 類型
缺陷
描述清晰度
基本清楚
活躍度
活躍
技術堆疊
java
領域
api, databases

研究方向

Start in Client.java at Client.lambda$insert$3, Client.insert, and runAsyncOperation, then trace HttpAPIClientHelper.executeRequest around retry handling. Reproduce the NoHttpResponseException path with the stated async-insert settings; done means a delivered insert can be retried without reusing an in-flight query ID or turning the transient fault into code 216.

由索引模型根據 Issue 內容生成。

描述

Description

Client.insert reuses requestSettings.getQueryId() across every retry attempt. After NoHttpResponseException (retried by default via ClientFaultCause.NoHttpResponse) the first request may already have reached the server and still be executing. The second attempt is then rejected with Code: 216 QUERY_WITH_SAME_ID_IS_ALREADY_RUNNING, which is not retryable, so a transient transport fault becomes a hard failure.

https://github.com/ClickHouse/clickhouse-java/pull/1721 introduced the retry and noted that "not every request may be retried safely". This is that case.

Steps to reproduce
  1. client.insert(table, data, format, settings) with default retry settings through a proxy or load balancer.
  2. The proxy closes the keep-alive connection after forwarding the request, while the server is still executing the insert (async_insert=1, wait_for_async_insert=1 makes the window at least the busy timeout).
  3. The client gets NoHttpResponseException, logs Retrying., and resends with the same query id. The server answers 216.
Error Log or Exception StackTrace
WARN Retrying. (com.clickhouse.client.api.Client)
org.apache.hc.core5.http.NoHttpResponseException: <host>:8443 failed to respond
    at org.apache.hc.core5.http.impl.io.DefaultBHttpClientConnection.receiveResponseHeader(DefaultBHttpClientConnection.java:333)
    ...
    at com.clickhouse.client.api.internal.HttpAPIClientHelper.executeRequest(HttpAPIClientHelper.java:453)
    at com.clickhouse.client.api.Client.lambda$insert$3(Client.java:1474)
    at com.clickhouse.client.api.Client.runAsyncOperation(Client.java:2012)
    at com.clickhouse.client.api.Client.insert(Client.java:1518)

com.clickhouse.client.api.ServerException: Code: 216. DB::Exception: Query with id = b266240e-d406-4874-ba3b-cd1b0bd969f8 is already running. (QUERY_WITH_SAME_ID_IS_ALREADY_RUNNING) (version 26.4.1.2359 (official build))
    at com.clickhouse.client.api.internal.HttpAPIClientHelper.readError(HttpAPIClientHelper.java:403)
    at com.clickhouse.client.api.internal.HttpAPIClientHelper.executeRequest(HttpAPIClientHelper.java:467)
    at com.clickhouse.client.api.Client.lambda$insert$3(Client.java:1474)
    at com.clickhouse.client.api.Client.runAsyncOperation(Client.java:2012)
    at com.clickhouse.client.api.Client.insert(Client.java:1518)
    at com.clickhouse.client.api.Client.insert(Client.java:1408)
Expected Behaviour

A retry after a fault where the first request may have been delivered does not collide with its own earlier attempt. Either the retry uses a fresh query id, or the client does not retry once the request body has been sent.

Configuration
Client Configuration

Defaults: retry=3, retryOnFailures = [NoHttpResponse, ConnectTimeout, ConnectionRequestTimeout, ServerRetryable]. Caller sets queryId and insert_deduplication_token per insert (clickhouse-kafka-connect v1.6.0).

Environment
  • Cloud
  • Client version: client-v2 0.9.5 (insert loop unchanged on 0.9.6, 0.10.0 and main)
  • Language version: Java 21.0.10
  • OS: Linux
ClickHouse Server
  • ClickHouse Server version: 26.4.1 and 26.6.1
  • Non-default settings: async_insert=1, wait_for_async_insert=1, async_insert_busy_timeout_ms=200

Related: https://github.com/ClickHouse/clickhouse-java/issues/1529 (same code, different cause). Downstream: https://github.com/ClickHouse/clickhouse-kafka-connect/issues/847.

主要語言
Java
星號
1.6k
分支
649
平均合併
1 天 20 小時
30 天內合併 PR
39

環境準備

從這裡開始

  1. 先讀完整個 Issue，再讀專案的貢獻指南。
  2. 在 Issue 下留言說明你要接手 —— 這能避免兩個人做同樣的事。
  3. Fork 儲存庫，在一個分支上完成修改。
  4. 送出 Pull Request，並在描述裡引用這個 Issue 編號。

ClickHouse/clickhouse-java 的其他 Issue

查看 ClickHouse/clickhouse-java 的全部 Issue

相似的 Issue

更多 Java Issue

把新 issue 寄到你的電子郵件信箱

精選適合新手參與的 GitHub issue 摘要。