[#963] Add txHash deduplication to indexer endpoints#971
Conversation
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub. |
project7-interns
left a comment
project7-interns
left a comment
There was a problem hiding this comment.
Verdict: REQUEST CHANGES
Summary
The DB-first dedup pattern is fine for single-row endpoints, but the trade endpoint can index multiple rows per tx. The new early-return check treats any existing tx_hash + token_address row as a fully cached success, which can mask partial indexing and drop remaining trade logs on retries.
Findings
- [high] Trade dedup can incorrectly short-circuit after a partial insert.
src/app/api/index/trade/route.tsnow returns{ ok: true, cached: true }if it finds any existing row for the submittedtx_hashandtoken_address. But the handler can upsert multiple trade rows for one tx (differentlog_indexvalues). If the first attempt inserted one row and then failed before indexing the rest, a retry would hit this dedup check and skip the remaining logs forever. That loses trade history instead of just avoiding duplicate work.- File:
src/app/api/index/trade/route.ts:29 - Suggestion: Dedup trade retries at the per-log level, or only return cached success when the receipt's expected relevant logs are all already present (for example by comparing the set/count of matching
log_indexvalues).
- File:
Decision
Requesting changes because the current trade dedup logic can turn a retriable partial failure into permanent missing indexed rows.
…g failures Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
project7-interns
left a comment
project7-interns
left a comment
There was a problem hiding this comment.
Verdict: APPROVE
Summary
The re-review update resolves the blocking trade-indexer issue by removing the coarse early dedup from src/app/api/index/trade/route.ts, so retries can still recover partial multi-log indexing failures. The remaining DB-first dedup checks are confined to the 1:1 endpoints where they correctly avoid redundant RPC/IPFS work.
Findings
- No blocking findings.
Decision
Approving because the previous correctness issue in the trade route is fixed and the dedup strategy now matches the data shape of each indexer endpoint. Checks visible to me were still pending at review time.
2 participants
Summary
200 { ok: true, cached: true }— skips all RPC/IPFS workFixes #963
Test plan
{ ok: true, cached: true }🤖 Generated with Claude Code