Cursor on macOS takes 12–23 s to quit ("Saving UI state", `ShutdownHangWatchdog ... quit exceeded budget`): `fseventsd` stuck at 100% CPU, fixed by restarting macOS

Context

Source(s): https://forum.cursor.com/t/slow-quit-20-23s-saving-ui-state-caused-by-enopro-filesystem-provider-race-on-close/174380 Signal: 62 views; the original poster's own resolution reply is the accepted answer, after following staff's diagnostic steps (October 2026). Status: reported working by source, not reproduced by us.

Problem

Quitting Cursor (Cmd+Q) consistently hangs on "Saving UI state" for 12–23 seconds. main.log shows ENOPRO: No file system provider found lines around the quit, which look like a shutdown race but are not the cause (see cause).

Error (redacted / normalized)

[ShutdownHangWatchdog] quit exceeded budget of 8000ms (elapsed=15698ms)
[ShutdownHangWatchdog] will-shutdown joiner 'applicationStorage' slow: elapsed=11911ms
[ShutdownHangWatchdog] will-shutdown joiner 'workspaceStorage' slow: elapsed=11911ms

[storage main] figuring out stats for slow DB on close() resulted in an error ENOPRO: No file system provider found for resource globalStorage/state.vscdb
CodeExpectedError: ENOPRO: No file system provider found for resource globalStorage/storage.json

No ptyHost heartbeat after 6 seconds

Applies when

Cursor 3.24.12 (first report) and 3.23.12 (the resolution reply) on macOS 26.6.2, Apple Silicon. The machine had been up for 47 days.

Confirm it is this problem

1. Quit Cursor, then open the newest folder under ~/Library/Application Support/Cursor/logs/ and search main.log for ShutdownHangWatchdog (staff). → expect: quit exceeded budget with applicationStorage / workspaceStorage joiners slow, as above.
2. Start Cursor from Terminal with open -a Cursor --args --disable-extensions and quit again (staff). → expect: still slow, so it is not an extension.
3. Check CPU use of fseventsd, e.g. in Activity Monitor or with top -o cpu (the tool is our inference; the original poster only reports the result). → expect: fseventsd near 100% CPU.

Not this when / boundary

fseventsd is idle and the quit is still slow with extensions disabled; then send the ShutdownHangWatchdog lines to Cursor support as staff asked (our inference). A slow quit caused by a multi-GB globalStorage/state.vscdb is a different problem; the original poster ruled it out after clearing a 4.4 GB database.

Root cause

Per the original poster's investigation: macOS fseventsd was using 100% CPU after 47 days of uptime, which slowed disk writes for every app, and Cursor's storage save at quit was stuck behind it. Staff explained that the ENOPRO lines are a side effect: the "slow DB on close" check only runs after closing has already taken more than a couple of seconds, by which time the file system layer is shutting down, so the check fails without adding to the wait.

Fix

1. Restart macOS (the original poster's fix). → expect: fseventsd back to normal CPU, and Cursor quits instantly, with and without extensions.
2. After quitting Cursor, check for leftover cursor-agent worker or MCP npm exec processes, e.g. pgrep -fl cursor-agent (the original poster found several still running after quit, one from an older Cursor version; the command is our inference). Quit them, or restart if they stay (our inference).

How to verify

Cmd+Q closes Cursor in a couple of seconds, and the next main.log has no quit exceeded budget line.

What did not work

Clearing a 4.4 GB state.vscdb, removing 159 stale workspaceStorage folders, disabling the remote-ssh and remote-containers extensions, and launching with --disable-extensions did not fix the quit time (the original poster).

Risks

The original poster saw the cursor-agent worker's API key passed on its command line (--api-key crsr_…), so any local user could read it with ps; they rotated the key. On a shared Mac, rotate the key if you saw it there, and never paste ps output or logs with a key into a public thread. Don't delete state.vscdb to "fix" a slow quit; it holds chat and agent data, and it was not the cause here (our inference).

Verified scope

Cursor 3.23.x–3.24.x on macOS 26.6, October 2026; the `fseventsd` cause is a macOS condition, not version-specific (our inference).

Becomes obsolete when

Cursor stops a slow disk from blocking quit, or stops passing the API key on the command line.