Aiven-Open / Aiven-Open/opensearch-connector-for-apache-kafka
Retry on conflict should have its own config
- 主要言語
- Java
- スター
- 95
- フォーク
- 61
- 平均マージ
- 11時間 52分
- マージ済み PR(30日)
- 7
説明
Hi, currently retry on conflict value is coupled to max.inflight.request, meaning that if we set max.inflight.request to 1, then the retry on conflict value will be 1. And it is capped at 3. Is there any reasoning why retry on conflict doesn't have their own config instead of coupling it with the inflight request and capped it to 3 ?
```
if (config.indexWriteMethod() == IndexWriteMethod.UPSERT) {
bulkOperation = BulkOperation.of(b -> b.update(u -> u.id(documentId)
.index(indexName)
.routing(routing)
.document(binaryData)
.upsert(binaryData)
.docAsUpsert(true)
.retryOnConflict(Math.min(config.maxInFlightRequests(), 3))));
```
I can help to make a PR if this is not intended.
Thank you
コントリビューションガイド
調査の方向性
Start at the UPSERT branch shown in the issue, then trace how max.inflight.request is defined and passed through the connector configuration. Done means retryOnConflict is controlled by its own configuration rather than Math.min(config.maxInFlightRequests(), 3), with the resulting behavior covered by the relevant project tests.
索引モデルが issue の本文から書いたものです。
評価
- 技術スタック
- java, kafka
- 領域
- backend, distributed-systems, search
- issue の種類
- 機能追加
- 難易度
- 3/5
- 見積もり時間
- 1〜2日
- 活発さ
- 静か
- 明瞭さ
- おおむね明確
- 初心者へのやさしさ
- 62/100