Connecting to Remote-SSH always fails

Support Bug Reports
1

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

Connecting a ubuntu server with my macbook always encounters this error:

2025-12-14 20:12:43.410 [info] (ssh_tunnel) stderr: ln: failed to create hard link ‘/tmp/cursor-remote-lock.6ca0869f6193ac0d40fa6ae74ed01260’: File exists

2025-12-14 20:12:43.449 [info] (ssh_tunnel) stdout: Install in progress, sleeping for a bit…

Steps to Reproduce

Connecting directly should reproduce this bug. Moreover, I discovered the issue seems to be that my Ubuntu server is a ‘worker’ provisioned by a cloud vendor; its operating system runs inside Docker, and its system disk (/root) is an NFS.

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

Version: 2.2.20
VSCode Version: 1.105.1
Commit: b3573281c4775bfc6bba466bf6563d3d498d1070
Date: 2025-12-12T06:29:26.017Z
Electron: 37.7.0
Chromium: 138.0.7204.251
Node.js: 22.20.0
V8: 13.8.258.32-electron.0
OS: Darwin arm64 25.0.0

Does this stop you from using Cursor

Yes - Cursor is unusable

SSH connection Error
Every time after the upgrade, Cursor Remote-SSH times out during remote server installation despite a successful download and extraction
3

I’ve retried often and sometimes the connection succeeds. This persistent issue has been around for a long time; I used retries as a temporary workaround. It consistently reappears after a Cursor upgrade. I attempted a manual cleanup using rm -r /tmp/cursor* before retrying, but this only solved the problem intermittently.

5

Here is the entire log from Remote-SSH when I’m encountering this bug.

cursor-remote-ssh-bug.txt (95.3 KB)

7

Hey, thanks for the report. I see two issues in the logs:

  1. Stuck lock file from previous installation
  2. Slow extraction on NFS disk - 30+ seconds exceeding timeout

Solution:

On the server, run:

killall bash
rm -rf ~/.cursor-server /tmp/cursor-remote-lock.* /tmp/cursor-server-*.tar.gz

Then in Cursor, add to Settings (JSON):

"remote.SSH.connectTimeout": 120

This will increase the timeout for your slow NFS environment.

If there’s a local disk on the server (check via df -h), you can speed up installation:

mkdir -p /tmp/cursor-server-cache
ln -s /tmp/cursor-server-cache ~/.cursor-server

NFS in your case means the server files are physically stored on remote cloud provider storage and accessed over network, so extracting thousands of small files is slower than on a local disk.

1 Like
8

Hi, thanks for your solution, this helped me a lot! :smiling_face_with_three_hearts:

1 Like