[BUG] Cowork VM connection timeout on Snapdragon X ARM64 — guest never connects after add_plan9_shares, persists through full reinstall

Status Closed — duplicate
Maintainer reply None cached
Activity 1 comment · opened Aug 11, 2026 · closed Aug 15, 2026

Preflight Checklist

  • [x] I have searched existing issues and this hasn't been reported yet
  • [x] This is a single bug report (please file separate reports for different bugs)
  • [x] I am using the latest version of Claude Code

What's Wrong?

Cowork fails to start on Windows 11 ARM64 (Snapdragon X Plus) with
"VM connection timeout after 60 seconds" on every attempt.

The failure is consistent and reproducible. The VM itself starts perfectly —
network creation, VM config, boot, and plan9 share attachment all complete in
under 500ms with no errors. The startup then stalls waiting for the guest
agent to connect back to the host, polls for 60 seconds, and times out.
The last completed step is always add_plan9_shares.

This is not a stale-state or corrupted-cache problem. I verified this by
deleting the entire vm_bundles directory, which forced a fresh initrd
download and a fresh copy of smol-bin.arm64.vhdx from the app package.
The fresh bundle downloads and copies successfully with no EBUSY lock,
the VM boots cleanly — and the guest still never connects. A complete
uninstall and reinstall of Claude Desktop produced the identical result.

Every host-side prerequisite has been verified as correct (see
Additional Information below). The host configuration is not the problem.

What Should Happen?

The Cowork workspace VM should start and the guest agent should establish
its connection to the host, allowing bash commands and code execution to run
in the isolated Linux environment.

Error Messages/Logs

=== Attempt 1 — initial failure ===

2026-08-11 15:04:21 [info] [VM:start] Copying smol-bin.arm64.vhdx to bundle:
  C:\Program Files\WindowsApps\Claude_1.26832.0.0_arm64__pzs8sxrjxfjjc\app\resources\smol-bin.arm64.vhdx
  -> C:\Users\<user>\AppData\Roaming\Claude\vm_bundles\claudevm.bundle\smol-bin.vhdx
2026-08-11 15:04:21 [info] [VM:start] smol-bin.arm64.vhdx copied successfully
2026-08-11 15:04:21 [info] [VM:start] Configuring Windows VM service...
2026-08-11 15:04:21 [info] [VM:start] Windows VM service configured
2026-08-11 15:04:21 [info] [VM:steps] create_network started
2026-08-11 15:04:21 [info] [VM:steps] create_network completed (0ms)
2026-08-11 15:04:21 [info] [VM:steps] create_vm_config started
2026-08-11 15:04:21 [info] [VM:steps] create_vm_config completed (88ms)
2026-08-11 15:04:22 [info] [VM:steps] vm_boot started
2026-08-11 15:04:22 [info] [VM:steps] vm_boot completed (157ms)
2026-08-11 15:04:22 [info] [VM:steps] add_plan9_shares started
2026-08-11 15:04:22 [info] [VM:steps] add_plan9_shares completed (10ms)
2026-08-11 15:04:29 [info] [startVM] VM startup already in progress, waiting...
2026-08-11 15:04:32 [info] [VM:start] Still waiting for guest connection... 10139ms elapsed, 21 polls
2026-08-11 15:04:42 [info] [VM:start] Still waiting for guest connection... 20306ms elapsed, 41 polls
2026-08-11 15:04:52 [info] [VM:start] Still waiting for guest connection... 30501ms elapsed, 61 polls
2026-08-11 15:05:03 [info] [VM:start] Still waiting for guest connection... 40659ms elapsed, 81 polls
2026-08-11 15:05:13 [info] [VM:start] Still waiting for guest connection... 50864ms elapsed, 101 polls
2026-08-11 15:05:22 [error] [VM:start] Connection timeout, last completed step: add_plan9_shares
2026-08-11 15:05:22 [info] Dispatching startup error: VM connection timeout after 60 seconds
2026-08-11 15:05:22 [error] [VM:start] Startup failed: Error: VM connection timeout after 60 seconds
    at Ut (C:\Program Files\WindowsApps\Claude_1.26832.0.0_arm64__pzs8sxrjxfjjc\app\resources\app.asar\.vite\build\index.chunk-sExhvzL4.js:10:5569) {
  [stack]: [Getter/Setter],
  [message]: 'VM connection timeout after 60 seconds'
}
2026-08-11 15:05:22 [info] [VM:start] Skipping auto-reinstall (already attempted once)


=== Retry — previous VM was never torn down, holding the disk ===

2026-08-11 15:06:33 [info] [KernelBugMonitor] State reset
2026-08-11 15:06:33 [info] [VM:start] Beginning startup, bundlePath=C:\Users\<user>\AppData\Roaming\Claude\vm_bundles\claudevm.bundle
2026-08-11 15:06:33 [info] [VM:start] Bundle version: 6d1538ba6fecc4e5c5583993c4b30bb1875f0f5a
2026-08-11 15:06:33 [info] [VM:start] VM instance ID: f17e2806-0716-421e-b179-725178114926
2026-08-11 15:06:33 [warn] [VM:start] smol-bin.vhdx locked by running VM (EBUSY), skipping copy
2026-08-11 15:06:43 [info] [VM:start] Still waiting for guest connection... 10191ms elapsed, 21 polls
2026-08-11 15:06:53 [info] [VM:start] Still waiting for guest connection... 20369ms elapsed, 41 polls


=== After deleting vm_bundles entirely — completely fresh download, no lock, SAME FAILURE ===

2026-08-11 15:49:00 [info] [warm] No warm file found for initrd at current version
2026-08-11 15:49:00 [info] [cache] No cached compressed file for initrd
2026-08-11 15:49:00 [info] initrd not found, downloading...
2026-08-11 15:49:02 [info] All files ready in C:\Users\<user>\AppData\Roaming\Claude\vm_bundles\claudevm.bundle
2026-08-11 15:49:02 [info] [VM:steps] download_and_sdk_prepare completed (61437ms)
2026-08-11 15:49:02 [info] [VM:steps] load_swift_api started
2026-08-11 15:49:02 [info] [VM:steps] load_swift_api completed (0ms)
2026-08-11 15:49:02 [info] [Callbacks] Initializing VM event callbacks...
2026-08-11 15:49:03 [info] [Callbacks] VM event callbacks initialized
2026-08-11 15:49:03 [info] [VM:start] smol-bin.arm64.vhdx copied successfully
2026-08-11 15:49:03 [info] [VM:start] Windows VM service configured
2026-08-11 15:49:03 [info] [VM:steps] create_network completed (18ms)
2026-08-11 15:49:03 [info] [VM:steps] create_vm_config completed (112ms)
2026-08-11 15:49:03 [info] [VM:steps] vm_boot completed (366ms)
2026-08-11 15:49:03 [info] [VM:steps] add_plan9_shares completed (6ms)
2026-08-11 15:49:14 [info] [VM:start] Still waiting for guest connection... 10166ms elapsed, 21 polls
2026-08-11 15:49:24 [info] [VM:start] Still waiting for guest connection... 20338ms elapsed, 41 polls
2026-08-11 15:49:34 [info] [VM:start] Still waiting for guest connection... 30520ms elapsed, 61 polls
2026-08-11 15:49:44 [info] [VM:start] Still waiting for guest connection... 40686ms elapsed, 81 polls

Steps to Reproduce

  1. Install Claude Desktop (arm64) on Windows 11 ARM64, Snapdragon X Plus
  2. Open the Cowork tab
  3. Error appears: "Failed to start Claude's workspace — VM connection timeout after 60 seconds"
  4. Persists across every restart, workspace reinstall, and full app reinstall

Claude Model

None

Is this a regression?

No, this never worked

Last Working Version

_No response_

Claude Code Version

1.26832.0.0

Platform

Other

Operating System

Windows

Terminal/Shell

PowerShell

Additional Information

This is a Claude Desktop Cowork issue, not Claude Code CLI.

=== Hardware / OS ===
Device: Microsoft Surface Pro 12
CPU: Snapdragon X Plus (8-core) @ 3.30 GHz, 8 logical processors
RAM: 15.6 GB
Storage: 236.8 GB SSD (UFS), 101.1 GB free
OS: Windows 11 Home, ARM64, OS Build 26200 (25H2)
(Get-ComputerInfo reports WindowsProductName as
"Windows 10 Home" / WindowsVersion 2009 — this is the known
legacy registry value on Windows 11, not an actual Win10 install)
Claude Desktop: 1.26832.0.0 (arm64)
Package: Claude_1.26832.0.0_arm64__pzs8sxrjxfjjc

=== Host prerequisites — ALL VERIFIED CORRECT ===
CPU virtualization (Task Manager): Enabled
VirtualMachinePlatform: Enabled
Microsoft-Hyper-V-All: Enabled
Microsoft-Hyper-V: Enabled
Microsoft-Hyper-V-Tools-All: Enabled
Microsoft-Hyper-V-Management-PowerShell: Enabled
Microsoft-Hyper-V-Services: Enabled
Microsoft-Hyper-V-Management-Clients: Enabled
Microsoft-Hyper-V-Hypervisor: Superseded (expected on this build)
HypervisorPlatform: Enabled
Microsoft-Windows-Subsystem-Linux: Enabled
CoworkVMService: Running
Free disk space: 101.1 GB

=== Troubleshooting already performed, all without effect ===

  1. Restarted Claude and rebooted the machine multiple times
  2. Restart-Service CoworkVMService -Force (service confirmed Running)
  3. wsl --shutdown to clear any stale VM
  4. Enabled VirtualMachinePlatform and Microsoft-Windows-Subsystem-Linux
  5. Killed all lingering VM worker processes
  6. Deleted %APPDATA%\Claude\vm_bundles entirely, forcing a fresh initrd

download and a fresh smol-bin.arm64.vhdx copy — confirmed successful,
no EBUSY lock on the retry

  1. Full uninstall and clean reinstall of Claude Desktop (arm64)

Every one of these produced the identical failure at the identical step.

=== Analysis ===
The host side is fully functional. The VM hypervisor works — the guest boots
in 366ms and plan9 shares attach in 6ms with no errors. The failure is
isolated to the guest agent inside the VM never establishing its connection
back to the host after add_plan9_shares completes.

Since a completely fresh bundle (new initrd, new vhdx, new app package)
reproduces this identically, the problem appears to be in the arm64 guest
image or its host transport, not in any user-side state or configuration.

Two secondary observations that may be relevant:

  • "Skipping auto-reinstall (already attempted once)" means the app gives up

on self-repair after one try and never retries, even across app restarts.

  • On retry after a failed start, the previous VM is not torn down and holds

smol-bin.vhdx (EBUSY), so subsequent attempts silently reuse a stale disk.
Cleaning this up is not exposed to users anywhere in the UI.

=== Related issues ===
#42784, #44007, #46516, #47737, #51277, #65128, #67066 — all appear to be
the same ARM64 failure mode, and all have been closed as duplicates without
a resolution being published.

<img width="365" height="269" alt="Image" src="https://github.com/user-attachments/assets/76709016-d8ee-4696-ac77-3ca916e90058" />
<img width="1097" height="684" alt="Image" src="https://github.com/user-attachments/assets/bb4e616c-cdd3-4b9e-a953-e282af258494" />
<img width="1084" height="542" alt="Image" src="https://github.com/user-attachments/assets/aef5bcff-8ce1-4991-a97f-000fe2b9d286" />
<img width="734" height="382" alt="Image" src="https://github.com/user-attachments/assets/b82edeff-e894-420f-b5ca-331b7fe7379e" />
<img width="326" height="85" alt="Image" src="https://github.com/user-attachments/assets/4ea1f537-d223-45f7-bff9-fe4c72b26d62" />
<img width="439" height="95" alt="Image" src="https://github.com/user-attachments/assets/8cbb34bb-3800-4984-9dda-48221ba50d3b" />
<img width="727" height="328" alt="Image" src="https://github.com/user-attachments/assets/59ef126a-1e2f-41b3-a044-cdc5bdfabce6" />

View original on GitHub ↗

This issue has 1 comment on GitHub. Read the full discussion on GitHub ↗