token #30

New issue

Closed

opened 2026-02-27 15:38:00 +03:00 by kerem · 1 comment

kerem commented

2026-02-27 15:38:00 +03:00

Owner

Originally created by @yulin0629 on GitHub (Feb 20, 2026).
Original GitHub issue: https://github.com/NikkeTryHard/zerogravity/issues/33

Rate limiter not reset when token is updated via POST /v1/token

Description

When switching Google accounts, injecting a new OAuth token via POST /v1/token does not reset the per-model rate limiter state. The proxy continues to reject requests based on the previous account's 429 status, even though the new account has full quota.

Steps to Reproduce

Use Account A until a model hits RESOURCE_EXHAUSTED (429)
Switch to Account B in Antigravity app
Inject new token: curl -X POST http://localhost:8741/v1/token -d '{"token":"ya29.new_token"}'
Send a request to the same model

Expected

Request succeeds (new account has quota).

Actual

Rate limiter rejected request: Model gemini-3-flash is rate-limited
(2x consecutive RESOURCE_EXHAUSTED). Retry in 10068s.

The rate limiter still holds the old account's 429 delay.

Workaround

docker compose down && docker compose up -d (full container recreate clears in-memory state). docker compose restart is not sufficient.

Suggested Fix

Reset all per-model rate limiter state when POST /v1/token receives a new token.

Version

v1.1.9 (ghcr.io/nikketryhard/zerogravity:v1.1.9), macOS Docker arm64.

Originally created by @yulin0629 on GitHub (Feb 20, 2026). Original GitHub issue: https://github.com/NikkeTryHard/zerogravity/issues/33 ## Rate limiter not reset when token is updated via POST /v1/token ### Description When switching Google accounts, injecting a new OAuth token via `POST /v1/token` does not reset the per-model rate limiter state. The proxy continues to reject requests based on the **previous** account's 429 status, even though the new account has full quota. ### Steps to Reproduce 1. Use Account A until a model hits `RESOURCE_EXHAUSTED` (429) 2. Switch to Account B in Antigravity app 3. Inject new token: `curl -X POST http://localhost:8741/v1/token -d '{"token":"ya29.new_token"}'` 4. Send a request to the same model ### Expected Request succeeds (new account has quota). ### Actual ``` Rate limiter rejected request: Model gemini-3-flash is rate-limited (2x consecutive RESOURCE_EXHAUSTED). Retry in 10068s. ``` The rate limiter still holds the old account's 429 delay. ### Workaround `docker compose down && docker compose up -d` (full container recreate clears in-memory state). `docker compose restart` is **not** sufficient. ### Suggested Fix Reset all per-model rate limiter state when `POST /v1/token` receives a new token. ### Version v1.1.9 (`ghcr.io/nikketryhard/zerogravity:v1.1.9`), macOS Docker arm64.

kerem closed this issue

2026-02-27 15:38:00 +03:00

kerem commented

2026-02-27 15:38:01 +03:00

Author

Owner

@NikkeTryHard commented on GitHub (Feb 20, 2026):

should be fixed for v1.2.0 please wait for stable release and test when it releases

@NikkeTryHard commented on GitHub (Feb 20, 2026): should be fixed for v1.2.0 please wait for stable release and test when it releases