fix: destroy the request as soon as the body exceeds the size cap so a flooding client cannot hold the sidecar open - #18
Merged
Conversation
Bug review of EauDoon/consequence-rail found: - HIGH: propose, reserveRecourse, issuePermit, and acceptRecoveryQualification all mutate the in-memory action record (or the actions Map) BEFORE the eventStore.append fires. transition() is documented as failure-atomic (it appends then mutates record.state), but the calling code overwrites that property for its own pre-state writes. A failing event append leaves the record with reservation / permit / recovery_preflight fields set, but no STATE_TRANSITION or ACTION_PROPOSED event. The action is now permanently wedged because the rail cannot retroactively insert the missing event. - HIGH: PR #15 fixed the same class of bug in authorize and the transition helper, but the four methods above were not updated. Fix: in each method, build the new field values into a local variable first, perform the eventStore.append (or transition, which appends), and only commit the field writes to the record after the append returns successfully. propose appends the ACTION_PROPOSED event before calling this.actions.set. reserveRecourse and issuePermit call transition() before writing record.reservation / record.permit. acceptRecoveryQualification appends the RECOVERY_PREFLIGHT_ACCEPTED event before writing record.recovery_preflight. Verified: node --test test/rail.test.js runs 83 tests, all pass.
…a flooding client cannot hold the sidecar open Bug review of EauDoon/consequence-rail found: - MEDIUM: src/http-server.js readJson kept iterating over the request stream after the cumulative byte count crossed MAX_BODY_BYTES, just clearing the local chunks buffer and continuing. The client kept sending bytes, the loop ran until the connection's natural EOF, and the sidecar only returned 413 after the entire oversized body had been transferred. A hostile client could pin the sidecar on a slow socket to extend a cheap bandwidth-exhaustion DoS. Fix: when the size cap is exceeded, call request.destroy() to terminate the underlying socket, then break out of the for-await loop. The 413 path then runs once with the partial body, no more bytes are accepted. The local chunks buffer is also no longer cleared on overflow because the socket is closed; this is the same final-state guarantee with less work in the loop. Verified: node --test test/rail.test.js test/recovery-preflight.test.js runs 90 tests, all pass.
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
There was a problem hiding this comment.
Devin Review found 1 potential issue.
1 flag not posted on this PR by your GitHub settings — view it in Devin Review. (Configure)
Comment on lines
+66
to
+67
| request.destroy(); | ||
| break; |
There was a problem hiding this comment.
🟡 Oversized streams lose error responses
When a streamed body exceeds the limit, request.destroy() closes its socket before the 413 response is sent. Clients receive an empty disconnect.
Prompt for agents
Preserve the early termination of oversized request uploads without destroying the socket before the HTTP error can be written. In src/http-server.js, readJson currently calls request.destroy(), then throws REQUEST_TOO_LARGE into the server catch handler. Since IncomingMessage.destroy() destroys the socket shared with ServerResponse, send() cannot deliver the 413. Rework this path so chunked or length-mismatched oversized requests stop consuming input and still produce the documented JSON 413 response. Add a raw-socket test that exceeds the limit without a declared oversized Content-Length and verifies both early termination and the response.
Was this helpful? React with 👍 or 👎 to provide feedback.
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.
Bug review of EauDoon/consequence-rail found:
Fix: when the size cap is exceeded, call request.destroy() to terminate the underlying socket, then break out of the for-await loop. The 413 path then runs once with the partial body, no more bytes are accepted. The local chunks buffer is also no longer cleared on overflow because the socket is closed; this is the same final-state guarantee with less work in the loop.
Verified: node --test test/rail.test.js test/recovery-preflight.test.js runs 90 tests, all pass.