Remote SSH: Agent execution provider deadline_exceeded; local editing works

[guyubinghu]

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

When using Cursor over Remote SSH (Windows client → SSH remote), Agent / Composer fails with:

• Title: Agent Execution Timed Out
• Detail: The agent execution provider did not respond in time. This may indicate the extension host is not running or is unresponsive.
• Error: ConnectError deadline_exceeded
• isRetryable: false

Normal remote editing and integrated terminal generally work. On the remote host, cursor-server is installed and processes are running (server-main, multiplex-server, extensionHost, fileWatcher, ptyHost). The machine has ample RAM (~90 GiB) and the workspace lives on a large mount (~2 TB); root overlay is not full.

The issue appears specific to Agent/Composer execution over Remote SSH, not to basic Remote SSH usability.

Steps to Reproduce

1. On Windows, open Cursor 2.6.21 (Stable).
2. Use Remote-SSH to connect to the development host (Ubuntu 22.04.5 LTS inside a Kubernetes pod; user account and IP as in Additional Information).
3. Open a large workspace on the mounted volume (e.g. ~/code/clb).
4. Open Composer / Agent and submit a request that triggers agent execution.
5. Observe: request fails with “Agent Execution Timed Out” and deadline_exceeded, with a Request ID shown in the error UI.

Expected Behavior

Agent / Composer should start successfully, connect to the execution provider in time, stream responses, and complete tool/execution steps over Remote SSH—similar to behavior on a local workspace.
error:
Request ID: 0cfc93c3-0578-4cde-8ae0-105aca443759
{“error”:“ERROR_CUSTOM”,“details”:{“title”:“Agent Execution Timed Out”,“detail”:“The agent execution provider did not respond in time. This may indicate the extension host is not running or is unresponsive.”,“isRetryable”:false,“shouldShowImmediateError”:true,“additionalInfo”:{},“buttons”:[{“label”:“Reload Window”,“reloadWindow”:{}}],“planChoices”:}}
Agent Execution Timed Out [deadline_exceeded]
ConnectError: [deadline_exceeded] Agent Execution Timed Out
at vscode-file://vscode-app/d:/Program%20Files/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:43805:12719
at async ZJy.createExecInstance (vscode-file://vscode-app/d:/Program%20Files/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:43805:10479)
at async xTA (vscode-file://vscode-app/d:/Program%20Files/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:34062:582888)
at async DJ.execute (vscode-file://vscode-app/d:/Program%20Files/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:34263:6603)
at async SRu.execute (vscode-file://vscode-app/d:/Program%20Files/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:43805:1323)
at async _Mw.execute (vscode-file://vscode-app/d:/Program%20Files/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:46878:2775)
at async _5u.buildComposerRequestContext (vscode-file://vscode-app/d:/Program%20Files/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:46888:3965)
at async _5u.streamFromAgentBackend (vscode-file://vscode-app/d:/Program%20Files/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:46888:5603)
at async _5u.getAgentStreamResponse (vscode-file://vscode-app/d:/Program%20Files/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:46888:13663)
at async vMe.submitChatMaybeAbortCurrent (vscode-file://vscode-app/d:/Program%20Files/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:34329:17597)
at async Ea (vscode-file://vscode-app/d:/Program%20Files/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:45813:4826)

Screenshots / Screen Recordings

Operating System

Windows 10/11

Version Information

IDE:
Version: 2.6.21 (system setup)
VSCode Version: 1.105.1
Commit: fea2f546c979a0a4ad1deab23552a43568807590
Date: 2026-03-21T22:09:10.098Z
Build Type: Stable
Release Track: Default
Electron: 39.8.1
Chromium: 142.0.7444.265
Node.js: 22.22.1
V8: 14.2.231.22-electron.0
OS: Windows_NT x64 10.0.26100

For AI issues: add Request ID with privacy disabled

Request ID: f9a7046a-279b-47e5-ab48-6e8dc12daba1

Additional Information

Remote environment (SSH target):

• Type: Kubernetes container pod
• Hostname: dg03podw9059165suhb
• OS: Ubuntu 22.04.5 LTS (container)
• Kernel: 5.14.0-284.11.1.el9_2.x86_64
• Pod IP: 10.234.88.135/19
• SSH user: W9059165
• Cursor remote install under home on mount /work/oppo_W9059165 (~2.0T); ~/.cursor-server present
• Remote cursor-server commit matches client: fea2f546c979a0a4ad1deab23552a43568807590

Other Request IDs (same issue class, earlier sessions):

• 5c2ab867-3786-40eb-b407-e4a0d3b1f056
• e8ef9b01-0cb9-47dc-9bff-d63f41806464
• 3cb24e23-5fb8-4d22-bcb9-6312c61f0d73

Already tried without fixing Agent timeout:

• SSH ServerAliveInterval / ServerAliveCountMax on client
• Removed ~/.cursor-server on remote and reconnected (fresh remote server install)
• Reload Window / restart Cursor
• Reduced extensions and parallel VS Code Remote usage

Stack trace path (client): D:\Program Files\cursor\ — workbench references createExecInstance, buildComposerRequestContext, streamFromAgentBackend / getAgentStreamResponse.

Background Agent: not used (no bc-… ID).

Does this stop you from using Cursor

Sometimes - I can sometimes use Cursor

[deanrie]

Hey, thanks for the detailed report. The environment info and request IDs help a lot.

This is a known issue. If you use Remote SSH with an NFS-mounted workspace like your ~2TB mount at /work/oppo_W9059165, the agent sandbox preflight check can time out because it relies on process.cwd(), which is slow on NFS. The extension host handler doesn’t register in time, so you get deadline_exceeded. The request never reaches our backend. It’s a client-side timeout.

Workaround to try: open your project from a local non-NFS path on the remote pod.

1. Copy the project to a local directory on the container, like /tmp/clb
2. Open that path in Cursor via Remote SSH

This avoids the NFS latency during the sandbox check.

A couple more things that can help:

• Kill any stale cursor-server processes on the remote:

  ps aux | grep cursor-server
  

  Kill old ones before reconnecting.

• Increase the SSH connect timeout in Cursor settings:

  "remote.SSH.connectTimeout": 120
  

The team knows about this SSH + NFS timeout issue. No ETA yet, but your report helps us prioritize.

Related thread with the same issue: Agent Execution Timed Out on Remote SSH — Everything Else Works Fine

Let me know if the local path workaround helps.