Repository navigation
fix(pm-api-latency): cohort to 5 native venues + venue dimension - #725
Merged
Merged
Conversation
Root cause: rank_matrix_query + uptime panels matched any venue label present in pmapi_* metrics — codex, codex-polymarket, codex-kalshi, predexon-*, mobula crept in via the pm-rate-limits harness scraping the same gauges. The leaderboard mixed real venues with paid relays and broken probes (success<70%), and surfaced bogus /products/<slug> pages. Fix: - Pin venue=~"polymarket|kalshi|limitless|manifold|myriad" on rank_matrix_query + uptime_24h/7d/30d + cold_connect_p50 panels. - Add dimension.venue so each native venue gets its own sub-leaderboard alongside the per-region tabs (Polymarket × us-east, etc.), matching the original PM cluster design.
This was referenced Jun 25, 2026
bakabhai993
pushed a commit
to bakabhai993/OpenChainBench
that referenced
this pull request
Sep 12, 2026
Zombie slugs (codex-polymarket, codex-kalshi, predexon-kalshi, predexon-limitless, predexon-polymarket) survived past the pre-collapse pm-api-latency split via stale materialize-worker snapshots. The pm-api-latency cohort fix in ChainBench#725 stops the bench from feeding them, but Upstash keeps the per-slug blobs around for TTL. Filter them out in buildProviders() so getProvider() returns undefined, the /products/<slug> route falls through to notFound(), the sitemap stops listing them, and related/compare candidates already filter them via the same set in related-providers.ts. Bumped buildProviders cache key v2→v3 to invalidate the ISR profile list immediately on deploy instead of waiting for the next benchmarks-tag revalidation.
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
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.
Two issues fixed:
1. Cohort pollution: rank_matrix_query and the uptime panels matched any venue label present in pmapi_* metrics. The pm-rate-limits harness scrapes the same gauges for codex, codex-polymarket, codex-kalshi, predexon-*, mobula → leaderboard mixed native venues with paid relays and broken probes (success<70%). It also surfaced /products/ pages for these junk slugs (codex-polymarket, predexon-kalshi, etc.) with successful HTTP 200.
Fix: pin
venue=~"polymarket|kalshi|limitless|manifold|myriad"on rank_matrix_query and on every uptime/cold-connect panel. Once the matrix excludes these slugs, /products/* pages for them stop having any benchmark appearance and 404 cleanly.2. No per-venue dimension: original PM cluster design called for sub-pages per venue via dimension.venue, mirroring dimension.region. Added it.
Requires worker Railway redeploy after merge so the snapshot picks up the new spec.