[Remote-Container Bug] After a reboot and reopening the last workspace, the container fails to reattach due to missing /run/user/1000/wayland-0 mount

[Roemer]

The problem I am facing is that my dev-containers automatically contain a wayland bind mount:

 {
        "Type": "bind",
        "Source": "/run/user/1000/wayland-0",
        "Destination": "/tmp/vscode-wayland-f52c3e63-d089-4a0f-aeda-ee401c2ec501.sock",
        "Mode": "",
        "RW": true,
        "Propagation": "rprivate"
},

which is fine by itself. But under some to me unknown circumstances it can happen after a complete reboot, that in WSL2 the wayland socket in /run/user/1000/ is missing (although it exists in /mnt/wslg/runtime-dir/). This causes the existing container to fail when I try to reopen vscode which had this mount previously attached. I then either have to reopen the folder in WSL and rebuild the container or actually shutdown WSL (wsl --shutdown) and restart it which usually makes the wayland socket reappear.

Not sure if this is actually something that can be done in vscode-remote or if this is a WSL issue. I will probably also open an issue in WSL.

• VSCode Version: 1.91.1
• Local OS Version: Windows 10
• Remote OS Version: WSL2/Debian
• Remote Extension/Connection Type: Containers over WSL

[chrmarti]

/run/user/1000/wayland-0 is a symlink to /mnt/wslg/runtime-dir/wayland-0 in my Ubuntu install. We currently use the symlink's path to mount the socket in the container. We could use the resolved path under /mnt if that is more robust.