Fix overview data integrity and freshness / 修复概览数据准确性与刷新问题 - #1111
Open
adibarra wants to merge 2 commits into
Open
Fix overview data integrity and freshness / 修复概览数据准确性与刷新问题#1111adibarra wants to merge 2 commits into
adibarra wants to merge 2 commits into
Conversation
Use producer-normalized total throughput once, minimize cost across eligible configurations, and retain measured endpoints that meet the minimum SLO. Separate measurement, run-attempt, and curve dates; expire client cache entries and guard refresh races. Correct physical-chip ingestion and add a conservative, dry-run-first topology repair command. Align dependent run/rankings pages. 中文：修复概览成本计算、SLO 数据覆盖与刷新问题。 直接使用生产端已按全部物理芯片归一化的总吞吐量，从满足最低 SLO 的有效配置中 选择最低成本。区分实测、运行与曲线快照日期，修复客户端缓存过期和刷新竞态。 校正摄取流程中的物理芯片数，并提供默认只读、保守判定的历史数据修复工具； 同步更新使用相同计算逻辑的运行详情页和排行榜。
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
This comment has been minimized.
This comment has been minimized.
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit 7bd081f. Configure here.
Read coach-mark and marker coordinates together in retryable assertions. Count successful updates toward the full-response backfill limit and scan past dateless profiles and no-op updates, with regression coverage. 中文：修复图表断言的时序问题，避免历史数据回填批次被无效记录占满。 在同一次可重试断言中读取提示箭头和数据点坐标；回填的数量限制只统计 实际更新的记录，跳过无法恢复时间戳或无需更新的记录，并补充回归测试。
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Note
High Risk
Changes core overview cost math, tier selection, ETL topology, and cache invalidation; published matrix and ranking numbers shift until optional DB backfills and cache invalidation are applied in production.
Overview
Corrects overview cost and coverage after an audit: total throughput is no longer penalized a second time for disaggregation, each hardware column picks the lowest-cost eligible FP4/FP8 stack, and endpoints above the selected minimum SLO count at their measured speed (with UI labels and updated methodology copy). Evidence tooltips now separate measurement completion, run-attempt, and curve snapshot dates.
Freshness: the overview BFF client cache gets a 5-minute TTL, periodic/ focus refresh without touching history, and bumped derived cache keys for overview,
/run, and/rankings.Pipeline: ingest maps fixed-sequence physical GPU counts like AgentX, normalizes colocated zero-decode Dynamo rows, stores
measurement_*_unix_secondsfrom profiles (preserved on re-ingest), documents throughput denominators in the public API, and addsdb:backfill-benchmark-topologyplus safer full-response date backfill/--limitbehavior. Rankings/run SEO text aligns with minimum-SLO semantics; Cypress and unit tests follow the new matrix values and refresh races.Reviewed by Cursor Bugbot for commit 93785bd. Bugbot is set up for automated code reviews on this repo. Configure here.