prometheus/prometheus

Remove OOO Head compaction's dependency on the in-order Head compaction

开放

#11,834 创建于 2023年1月10日

 (30 条评论) (0 个反应) (2 位负责人)Go (10,408 个派生)batch import
component/tsdbgood first issuehelp wanted

仓库指标

星标
 (64,042 个星标)
PR 合并指标
 (平均合并 11天 5小时) (30 天内合并 118 个 PR)

描述

Proposal

It is possible that the in-order Head stopped getting data for a while, while the OOO Head kept getting a lot of old data. In which case, we should compact the OOO data eventually.

Currently, we attempt OOO compaction only if the in-order Head gets compacted. We should remove that dependency to attempt OOO compaction every 2h as well. Irrespective of block duration here, we can attempt OOO compaction every 2h because there is no cap on how much OOO data you can ingest.

贡献者指南