Nvim always logs server_start:164 error in WSL (SOLVED: permissions issue)

[xudyang1]

Problem

In windows subsystem for linux (WSL), use neovim with custom init.lua or with --clean flag always produce the following error:

WRN 2023-11-15T16:47:33.489 ?.5332     server_start:164: Failed to start server: permission denied: /run/user/1000//
WRN 2023-11-15T16:48:25.884 ?.5564     server_start:164: Failed to start server: permission denied: /run/user/1000//
WRN 2023-11-15T16:49:03.825 ?.5888     server_start:164: Failed to start server: permission denied: /run/user/1000//
WRN 2023-11-15T16:49:17.203 ?.6133     server_start:164: Failed to start server: permission denied: /run/user/1000//
WRN 2023-11-15T16:49:31.533 ?.6262     server_start:164: Failed to start server: permission denied: /run/user/1000//

nvim or nvim --clean appends a line WRN .... to the file in ~/.local/state/nvim/log

Steps to reproduce

nvim --clean   # `:q`
cat ~/.local/state/nvim/log

Expected behavior

I also use neovim in powershell from the host system (windows 10), the log file (~/AppData/Local/nvim-data/log) does not contain such errors. Therefore, this may be a bug in neovim when used inside WSL environment.

Neovim version (nvim -v)

0.9.4

Vim (not Nvim) behaves the same?

i don't use vim

Operating system/version

ubuntu 22.04 (WSL)

Terminal name/version

windows terminal/ 1.19.2682.0

$TERM environment variable

xterm-256color

Installation

appimage

[justinmk]

permission denied: /run/user/1000//

This implies that the temp directory being calculated by vim_mktempdir is not
correct. Maybe this function is failing:

neovim/src/nvim/os/users.c

Lines 125 to 133 in 400b784

	int os_get_username(char *s, size_t len)
	{
	#ifdef UNIX
	return os_get_uname((uv_uid_t)getuid(), s, len);
	#else
	// TODO(equalsraf): Windows GetUserName()
	return os_get_uname((uv_uid_t)0, s, len);
	#endif
	}

• Is /run/user/1000/ an existing directory on your system?
• What does does this show:

  :echo v:servername
  :echo stdpath('run')
  :echo tempname()
  

• What happens if you start Nvim with the $TMPDIR env var set to the current directory?

  TMPDIR=. nvim
  

  • What does :echo tempname(), etc., say in that case?
• If you are interested in debugging this, should be easy to set a breakpoint here:

  neovim/src/nvim/fileio.c

  Line 3273 in 400b784

  static void vim_mktempdir(void)

[xudyang1]

permission denied: /run/user/1000//

This implies that the temp directory being calculated by vim_mktempdir is not correct. Maybe this function is failing:
...

Sorry for replying so late.

1. /run/user/1000 is an existing directory but it is empty and requires sudo.

2. Here are the results:

:echo v:servername

:echo stdpath('run') 
/run/user/1000/

:echo tempname()
/tmp/nvim.myUserName/0EKFgk/0

3. TMPDIR=. nvim results in :echo tempname() being directory ./nvim.myUserName/someRandomString/ under current directory when using nvim. After exiting neovim, only ./nvim.myUserName/ exist.
   Outputs of :echo v:servername and :echo stdpath('run') remain unchanged.

[xudyang1]

@justinmk I think this issue is particularly related to the failed service user@1000.service in WSL2 with systemd enabled.

$ systemctl --failed
  UNIT              LOAD   ACTIVE SUB    DESCRIPTION
● user@1000.service loaded failed failed User Manager for UID 1000

LOAD   = Reflects whether the unit definition was properly loaded.
ACTIVE = The high-level unit activation state, i.e. generalization of SUB.
SUB    = The low-level unit activation state, values depend on unit type.
1 loaded units listed.

$ systemctl status user@1000.service
× user@1000.service - User Manager for UID 1000
     Loaded: loaded (/lib/systemd/system/user@.service; static)
    Drop-In: /usr/lib/systemd/system/user@.service.d
             └─timeout.conf
     Active: failed (Result: exit-code) since Fri 2024-01-26 12:20:01 EST; 23min ago
       Docs: man:user@.service(5)
    Process: 1625 ExecStart=/lib/systemd/systemd --user (code=exited, status=1/FAILURE)
   Main PID: 1625 (code=exited, status=1/FAILURE)
        CPU: 15ms

Jan 26 12:20:01 DESKTOP-********* systemd[1]: Starting User Manager for UID 1000...
Jan 26 12:20:01 DESKTOP-********* systemd[1625]: pam_unix(systemd-user:session): session opened for user bell(uid=1000) by (uid=0)
Jan 26 12:20:01 DESKTOP-********* systemd[1625]: pam_systemd(systemd-user:session): Runtime directory '/run/user/1000' is not owned by UI>
Jan 26 12:20:01 DESKTOP-********* systemd[1625]: pam_systemd(systemd-user:session): Not setting $XDG_RUNTIME_DIR, as the directory is not>
Jan 26 12:20:01 DESKTOP-********* systemd[1625]: Trying to run as user instance, but $XDG_RUNTIME_DIR is not set.
Jan 26 12:20:01 DESKTOP-********* systemd[1]: user@1000.service: Main process exited, code=exited, status=1/FAILURE
Jan 26 12:20:01 DESKTOP-********* systemd[1]: user@1000.service: Failed with result 'exit-code'.
Jan 26 12:20:01 DESKTOP-********* systemd[1]: Failed to start User Manager for UID 1000.

$ sudo systemctl restart user@1000.service
Job for user@1000.service failed because the control process exited with error code.
See "systemctl status user@1000.service" and "journalctl -xeu user@1000.service" for details.

$ journalctl -xeu user@1000.service
Jan 26 12:47:44 DESKTOP-********* systemd[1]: Starting User Manager for UID 1000...
░░ Subject: A start job for unit user@1000.service has begun execution
░░ Defined-By: systemd
░░ Support: http://www.ubuntu.com/support
░░
░░ A start job for unit user@1000.service has begun execution.
░░
░░ The job identifier is 573.
Jan 26 12:47:44 DESKTOP-********* systemd[2356]: pam_unix(systemd-user:session): session opened for user bell(uid=1000) by (uid=0)
Jan 26 12:47:44 DESKTOP-********* systemd[2356]: pam_systemd(systemd-user:session): Runtime directory '/run/user/1000' is not owned by UI>
Jan 26 12:47:44 DESKTOP-********* systemd[2356]: pam_systemd(systemd-user:session): Not setting $XDG_RUNTIME_DIR, as the directory is not>
Jan 26 12:47:44 DESKTOP-********* systemd[2356]: Trying to run as user instance, but $XDG_RUNTIME_DIR is not set.
Jan 26 12:47:44 DESKTOP-********* systemd[1]: user@1000.service: Main process exited, code=exited, status=1/FAILURE
░░ Subject: Unit process exited
░░ Defined-By: systemd
░░ Support: http://www.ubuntu.com/support
░░
░░ An ExecStart= process belonging to unit user@1000.service has exited.
░░
░░ The process' exit code is 'exited' and its exit status is 1.
Jan 26 12:47:44 DESKTOP-********* systemd[1]: user@1000.service: Failed with result 'exit-code'.
░░ Subject: Unit failed
░░ Defined-By: systemd
░░ Support: http://www.ubuntu.com/support
░░
░░ The unit user@1000.service has entered the 'failed' state with result 'exit-code'.
Jan 26 12:47:44 DESKTOP-********* systemd[1]: Failed to start User Manager for UID 1000.
░░ Subject: A start job for unit user@1000.service has failed
░░ Defined-By: systemd
░░ Support: http://www.ubuntu.com/support
░░
░░ A start job for unit user@1000.service has finished with a failure.
░░
░░ The job identifier is 573 and the job result is failed.

• Also note that WSL Ubuntu-22.04 does not set $TMP, TMPDIR, $TEMP, and $USERPROFILE:

$ echo $TMP,$TMPDIR,$TEMP,$USERPROFILE
,,,

• But $XDG_RUNTIME_DIR is set despite the error message from the command $ systemctl status user@1000.service

$ echo $XDG_RUNTIME_DIR
/run/user/1000/

• /run/user/1000/ is owned by root user only, this may be the consequence of the failed service user@1000.service.

$ sudo ls -ld /run/user/1000/
drwx------ 2 root root 4096 Nov 15 10:49 /run/user/1000/

Related:

• WSL2 /run/user/1000 permission denied
• WSL2 /run/user/1000 owned by root
• disabling wslg makes $XDG_RUNTIME_DIR not user accessible

[xudyang1]

It could be this podman issue, solved by changing owner to current user sudo chown username:usergroup /run/user/1000.

Closed.