billflare.

Source-linked reports: Amounts and outcomes come from the original reports and linked replies.

D1operonatlas-d1-bulk-import-writes

OperonAtlas: a one-time D1 import wrote 320.76 million billable rows, about $321; the $1 alert came after.

Community user user33412 runs OperonAtlas, a non-commercial academic research database on Workers and D1, and says a one-time large database import and indexing wrote about 370.76 million D1 rows, 320.76 million billable, about $321. A $1 budget alert only arrived after the operation had finished. The author first says a support case requesting a one-time adjustment was opened, then says no billing option could be found; an MVP asked which was the case.

What happened

Reported cause
Write volume
Incident date
Date not recorded
First reported
2026-08-25
Billing outcome
Not recorded
Case progress
Not established
Public Cloudflare reply
Not linked

Original sources

Each entry links to the report or reply it describes.

Case timeline

  1. Author asks for an adjustment

    Open original source

    The author reports about $321 from a one-time import and indexing and a $1 budget alert that arrived only afterwards.

Community resourceWorkers +3

Cloudflare Cost Guard: a read-only agent skill for runaway-billing audits

Layer Cache’s open-source agent skill, announced by Domenico (@purposelyblnk), audits a Cloudflare codebase from source alone for paths that can run up uncontrolled billable usage, and reports findings without changing code or calling Cloudflare.

Workers

RetainDB: a reported $36k bill from a queue loop with 81 users.

Reddit user alameenswe, building RetainDB on Workers, KV, Durable Objects and Queues, reports a Cloudflare bill of about $36k for one month with 81 users; the post body says $35,000. He attributes it to a queue consumer that re-queued async writes, 12 unbatched Durable Object writes per memory write and a kv.list() authentication fallback.

Official docsAccount-wide usage-based products +3

Optimize costs

Cloudflare lists ways to find and reduce usage-based charges by product.