Skip to content

fix: continuation loop never stops (stream.switches is never incremented) #2187

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Cadenadb wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix: continuation loop never stops (stream.switches is never incremented) #2187

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
21 changes: 19 additions & 2 deletions app/routes/api.chat.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,6 +75,18 @@ async function chatAction({ context, request }: ActionFunctionArgs) {

const stream = new SwitchableStream();

/*
* FIX (2026-08-04): `stream.switches` is never incremented anywhere in this file
* (SwitchableStream only increments `_switches` inside `switchSource()`, which is
* never called here). That means the `stream.switches >= MAX_RESPONSE_SEGMENTS`
* safety check below never trips, and the "continue on length limit" branch can
* recurse forever for any model/provider that keeps returning finishReason "length"
* (observed in production: infinite loop, one real LLM call every ~40s, forever,
* silently consuming API credits). Tracking our own counter here restores the
* intended cap of MAX_RESPONSE_SEGMENTS.
*/
let continuationSegments = 0;

const cumulativeUsage = {
completionTokens: 0,
promptTokens: 0,
Expand Down Expand Up @@ -249,11 +261,16 @@ async function chatAction({ context, request }: ActionFunctionArgs) {
return;
}

if (stream.switches >= MAX_RESPONSE_SEGMENTS) {
if (continuationSegments >= MAX_RESPONSE_SEGMENTS) {
logger.error(
`Reached max token limit (${MAX_TOKENS}) and used all ${MAX_RESPONSE_SEGMENTS} continuation segments -- stopping instead of looping forever.`,
);
throw Error('Cannot continue message: Maximum segments reached');
}

const switchesLeft = MAX_RESPONSE_SEGMENTS - stream.switches;
continuationSegments++;

const switchesLeft = MAX_RESPONSE_SEGMENTS - continuationSegments;

logger.info(`Reached max token limit (${MAX_TOKENS}): Continuing message (${switchesLeft} switches left)`);

Expand Down
Loading