Ripcord

A dead-simple parachute for your server's disk.

The Problem

A full disk can crash a server and corrupt data. When this happens, you are locked out and cannot fix the problem.

The Solution

Ripcord runs once on a new system's first boot. It creates a single, large "ballast" file of uncompressible random data.

In an emergency, you delete this one file. This instantly frees up critical disk space, giving you time to log in and properly resolve the issue.

In an Emergency: Pull The Ripcord

To prevent a disk-full event and recover your system, delete this file:

# This is it. This is the command that saves the server.
rm /RIPCORD_EMERGENCY_DELETE_FOR_DISK_SPACE.dat

This single action gives you the breathing room you need.

Installation

Installation is a two-step process: place the script, then enable the service for your OS.

1. Place the Script (All Systems)

Copy ripcord.sh to /usr/local/sbin/ and make it executable.

sudo cp ripcord.sh /usr/local/sbin/ripcord.sh
sudo chmod +x /usr/local/sbin/ripcord.sh

2. Enable the First-Boot Service

Choose the instructions for your operating system.

A) For Debian / Ubuntu / RHEL (systemd)

# Create and enable the systemd service. It will run once on the next boot.
cat << EOF | sudo tee /etc/systemd/system/ripcord.service
[Unit]
Description=Ripcord Emergency Disk Space Reserve
[Service]
Type=oneshot
ExecStart=/usr/local/sbin/ripcord.sh
ExecStartPost=/bin/sh -c "systemctl disable ripcord.service"
[Install]
WantedBy=multi-user.target
EOF

sudo systemctl enable ripcord.service

B) For FreeBSD (rc.d)

# Create the rc.d script.
cat << EOF | sudo tee /usr/local/etc/rc.d/ripcord
#!/bin/sh
# PROVIDE: ripcord
# REQUIRE: LOGIN
. /etc/rc.subr
name="ripcord"
start_cmd="${name}_start"
stop_cmd=":"
ripcord_start() {
    if [ ! -f "/var/log/ripcord-deployed.flag" ]; then
        /usr/local/sbin/ripcord.sh
    fi
}
load_rc_config \$name
run_rc_command "\$1"
EOF

# Make it executable and enable it in rc.conf.
sudo chmod +x /usr/local/etc/rc.d/ripcord
sudo sysrc ripcord_enable="YES"

C) For OpenBSD (rc.local)

# Add the execution command to /etc/rc.local.
# This block will be checked on every boot, but only run once.
cat << EOF | sudo tee -a /etc/rc.local

# Deploy Ripcord on first boot if not already done
if [ ! -f "/var/log/ripcord-deployed.flag" ] && [ -x "/usr/local/sbin/ripcord.sh" ]; then
    /usr/local/sbin/ripcord.sh
fi
EOF

Configuration

To change the size percentage or filename, edit the variables at the top of the ripcord.sh script.

# Percentage of the total root filesystem size to use.
PERCENTAGE=5

# The emergency file. The name itself is the instruction.
TARGET_FILE="/RIPCORD_EMERGENCY_DELETE_FOR_DISK_SPACE.dat"

2.8 KiB Raw Blame History