StateManagerDisk silently resets corrupted persisted state

Author: harsh21234iCreated Sep 16, 2026Updated Sep 16, 2026

Describe the bug

StateManagerDisk.load_state() catches every deserialization exception and returns None without logging or distinguishing corruption from a missing state file. get_state() then treats the corrupted file as a brand-new state and can later overwrite the persisted data with defaults, causing silent backend state loss.

To Reproduce

python

import asyncio
import tempfile
from pathlib import Path

from reflex.istate.manager.disk import StateManagerDisk
from reflex.istate.manager.token import StateToken


async def main():
    with tempfile.TemporaryDirectory() as tmp:
        manager = StateManagerDisk()
        manager.__dict__["states_directory"] = Path(tmp)
        token = StateToken(ident="client", cls=int)

        manager.token_path(token).write_bytes(b"not-a-pickle")

        print("load_state:", await manager.load_state(token))
        print("get_state:", await manager.get_state(token))

        await manager.close()


asyncio.run(main())

Source: reflex-dev/reflex

View original on GitHub View discussion on GitHub