👻 The Server tmpfs Trap: /run/sshd Disappears After Reboot, SSH Refuses Connection

The Symptom

Server reboots. SSH hangs at connection. Console in: sshd running, port open, firewall fine. Still won't connect.

Check sshd logs: Missing privilege separation directory: /run/sshd

Why This Happens

Modern SSH uses privilege separation — sshd splits into privileged and unprivileged processes. The unprivileged child needs /run/sshd. But /run is tmpfs — it lives in RAM and gets wiped on every reboot. On minimal VPS installs, the systemd rule that recreates it can be missing.

Immediate Fix

mkdir -p /run/sshd
chmod 0755 /run/sshd
systemctl restart sshd

SSH works immediately. But the directory disappears again on next reboot.

Permanent Fix

# Create /etc/tmpfiles.d/sshd.conf
echo "d /run/sshd 0755 root root" > /etc/tmpfiles.d/sshd.conf

Verify: systemd-tmpfiles --create /etc/tmpfiles.d/sshd.conf

Lesson

Anything in /run or /tmp that a service depends on needs a tmpfiles.d rule. Never assume a directory survives a reboot just because it exists right now. Always verify your setup survives a reboot before calling it production-ready.