dsh web --stop / dsh web --restart — stop orphan dsh web processes cleanly

[moonquake2004]

Problem

When a terminal running dsh web is closed, the node process listening on port 3080 survives as an orphan. There is no built-in way to stop it:

• dsh has no stop / restart subcommand
• Ctrl+C in the original terminal only kills the foreground shell — the dsh web process (often backgrounded or launched via npx) keeps running
• lsof -i :3080 + kill works but is manual and undiscoverable for new users
• Port 3080 stays occupied, so starting dsh web again fails with "port in use" until the user finds and kills the old process

This is the most common friction point I've seen in dsh-doctor diagnostics (#2752 white-screen bugs often coincided with orphan processes) and in community reports.

Proposal

Three additions to the dsh web command:

1. dsh web --stop

Kill the process listening on the dsh web port (default 3080). Graceful SIGTERM with 5s timeout, then SIGKILL. Print what was killed (PID, uptime).

$ dsh web --stop
Stopping dsh web (PID 27158, uptime 3h12m)...
✓ Stopped (port 3080 free)

2. dsh web --restart

Equivalent to --stop + normal start. The most common intent when a user wants to "refresh" dsh.

$ dsh web --restart
Stopping dsh web (PID 27158)...
✓ Stopped
Starting dsh web on 127.0.0.1:3080...
✓ Ready

3. Auto-reclaim on dsh web start (optional)

When dsh web finds port 3080 occupied by another dsh instance (same profile), log a warning and auto-stop it before starting. This turns a confusing "port in use" error into a self-healing experience. Only auto-reclaim if the occupier is a dsh process (same profile path / dsh binary signature) — never kill an unknown occupier without asking.

$ dsh web
Port 3080 occupied by dsh web (PID 27158, same profile). Auto-stopping...
✓ Stopped old instance
Starting dsh web on 127.0.0.1:3080...
✓ Ready

Why this matters

• Discoverability: new users don't know they need lsof + kill — they just see "port in use" and restart their machine
• Consistency: pm2, Docker, and every daemon manager have stop/restart. dsh is the only one that doesn't
• Diagnostic accuracy: orphan processes cause stale state (old plugins loaded, wrong settings panel) that dsh-doctor can't fix without a restart the user doesn't know how to do

Current workaround (what I ship in my dotfiles)

dsh-reset() {
  local port="${1:-3080}"
  local pids=$(lsof -i :$port -sTCP:LISTEN -F p -n | sed 's/^p//')
  [ -n "$pids" ] && kill $pids && sleep 2 && kill -9 $pids 2>/dev/null
  dsh web --host 127.0.0.1 --port $port &
  disown
}

Works, but every user has to write this themselves. Upstream is the right place.

[AkiyamaKunka]

Meet the same issue when upgrading plugins.

[DonnieTangzhili]

https://github.com/DonnieTangzhili/dsh-ops-console可以试试我这个插件，希望可以帮助到你

[AkiyamaKunka]

404

[DonnieTangzhili]

https://github.com/DonnieTangzhili/dsh-ops-console