Prevent Linux OOM Killer From Crashing MySQL and Node Services

When the Linux kernel runs completely out of physical RAM blocks, it invokes a safety mechanism called the Out-Of-Memory (OOM) Killer. The OOM Killer scans active system processes, assigns an internal badness score (oom_score), and violently terminates the heaviest process (frequently your production database or web service) to prevent a total OS kernel panic.

If your application logs suddenly vanish or display Exit code 137 alongside kernel system alerts (dmesg | grep -i oom), your environment is suffering from unbuffered memory exhaustion.

The Dual-Layer Stability Remediation

To secure a low-resource or highly volatile server from unexpected runtime terminations, your engineering strategy must deploy two specific safety buffers:

1. Virtual Memory Page Spooling (Swap Allocation): If your server has no Swap space allocated, the absolute moment RAM usage touches 100%, services drop. Provisioning a dedicated Swap file onto the underlying SSD acts as a non-volatile safety overflow valve, absorbing sudden memory spikes.
2. Process Score Insulation (oom_score_adj): You can explicitly direct the Linux kernel's priority matrix to prioritize or completely protect critical services (like sshd or mysqld) from being evaluated as high-priority execution targets during low-memory panics.

The automated infrastructure script below provisions an optimized 4GB system Swap space layer and details how to immunize mission-critical system daemons.

Bash Linux Memory Optimization Script

#!/bin/bash

============================================================

BLUEPRINT 1: AUTOMATED STABLE SWAP SPACE PROVISIONING ENGINE

============================================================

Must be executed with root/sudo privileges: sudo ./allocate-swap.sh

SWAP_PATH="/swapfile" SWAP_SIZE_GB=4

Check if a Swap layer is already initialized on this machine

if swapon --show | grep -q "$SWAP_PATH"; then echo "[Status] Swap partition already active at $SWAP_PATH. Exiting allocation." exit 0 fi

echo "[Provisioning Storage] Creating a ${SWAP_SIZE_GB}GB data allocation block..."

fallocate reserves system blocks instantly without overhead write stress

sudo fallocate -l "${SWAP_SIZE_GB}G" "$SWAP_PATH"

if [ $? -ne 0 ]; then echo "Fallocate failed, falling back to dd block copier..." sudo dd if=/dev/zero of="$SWAP_PATH" bs=1M count=$((SWAP_SIZE_GB * 1024)) fi

echo "[Securing Permissions] Locking file access down to system root user..."

Strict permission structures are vital; unprivileged users must never read Swap memory

sudo chmod 600 "$SWAP_PATH"

echo "[Formating Schema] Mapping raw storage block into Swap space filesystem..." sudo mkswap "$SWAP_PATH"

echo "[Activating Subsystem] Mounting Swap space engine live into runtime context..." sudo swapon "$SWAP_PATH"

echo "[Enabling Persistence] Writing configuration parameters into system filesystem table..." if ! grep -q "$SWAP_PATH" /etc/fstab; then echo "$SWAP_PATH none swap sw 0 0" | sudo tee -a /etc/fstab fi

Set swappiness value. Lower values (e.g., 10) tell Linux to use swap only as emergency overflow

echo "[Tuning Swappiness] Adjusting VM execution trends..." sudo sysctl vm.swappiness=10 echo "vm.swappiness=10" | sudo tee -a /etc/sysctl.conf

echo "Success: Emergency virtual memory safety layer fully operational."

========================================================

BLUEPRINT 2: PROCESS OVERRIDE IMMUNIZATION CONFIGURATION

========================================================

To completely protect an essential process like MySQL from being killed by OOM,

you can modify its systemd unit file adjustment properties.

Run: sudo systemctl edit mysql

Inject the following block into the system override display interface:

read -r -d '' SYSTEMD_OVERRIDE_BLUEPRINT << 'EOF' [Service]

Set the OOM score adjustment matrix to the absolute minimum (-1000)

This completely exempts the service from the OOM Killer targeting pool.

OOMScoreAdjust=-1000 EOF