The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
docs	Skipped		Apr 27, 2026 7:02pm

@BugBot review

PR Summary

Medium Risk
Touches recurring retention cleanup jobs and the shared delete helper; incorrect chunking/limits could leave data unpruned or over-delete if predicates are wrong, though changes are mostly mechanical and bounded by existing retention filters.

Overview
Improves retention cleanup reliability at scale by chunking large workspace/ID lists before running SELECT/DELETE loops, avoiding oversized IN (...) queries and statement timeouts.

Introduces selectRowsByIdChunks and a generalized chunkedBatchDelete helper (with per-chunk batching, an optional per-batch side-effect hook, and an overall row cap), and updates the cleanup-logs, cleanup-soft-deletes, and cleanup-tasks jobs to use these chunked primitives; cleanup-logs also consolidates workflow-log deletion + file cleanup via the new hook and switches job log deletion to the shared wrapper.

^{Reviewed by Cursor Bugbot for commit 3c48b24. Bugbot is set up for automated code reviews on this repo. Configure here.}

Greptile Summary

This PR refactors the retention cleanup jobs to chunk large workspace-ID IN lists before issuing SELECT/DELETE queries, avoiding Postgres statement timeouts caused by probing every workspace range in the composite index. The duplicated per-table loop logic is consolidated into two shared helpers — chunkedBatchDelete (SELECT → optional side-effect → DELETE by ID) and selectRowsByWorkspaceChunks (SELECT-only, with a row cap).

Confidence Score: 4/5

Safe to merge — the fix addresses a real production failure with correct logic; the two P2 items are about failure-counter semantics and an uncapped per-run row budget that should be validated before deploying at scale.

No P0/P1 bugs found. Two P2 issues: result.failed semantics silently changed from row count to batch count (affects monitoring readability), and chunkedBatchDelete no longer has a fixed global row cap (could dramatically increase per-run work for large free-tier installs). All P2s, so ceiling is 4/5.

apps/sim/lib/cleanup/batch-delete.ts — the new chunkedBatchDelete carries the changed failed semantics and unbounded-per-chunk scaling; worth a close look before deploying to a large free-tier instance.

Important Files Changed

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[workspaceIds] --> B[chunkArray\n50 IDs per chunk]
    B --> C{for each chunk}

    C --> D[selectChunk\nSELECT up to batchSize rows]
    D --> E{rows.length == 0?}
    E -- yes --> F[next chunk]
    E -- no --> G[onBatch?\ne.g. delete S3 files]
    G --> H[DELETE rows by ID]
    H --> I{rows.length == batchSize\nAND batches < maxBatches?}
    I -- yes --> D
    I -- no --> F
    F --> C

    D -. error .-> ERR[result.failed++\nskip to next chunk]
    G -. error .-> ERR
    H -. error .-> ERR

    C -- all chunks done --> LOG[Log: total deleted\nchunk failures]

_{Reviews (1): Last reviewed commit: "fix(retention-job): add chunking strateg..." | Re-trigger Greptile}