VPS Out of Memory OOM Solutions - Linux Memory Optimization Guide

The server suddenly slows down, processes get killed for no apparent reason, MySQL goes down without warning... If you've run into these situations, there's a good chance that insufficient memory has triggered the OOM Killer. This article teaches you how to diagnose and solve VPS out-of-memory problems.

What Is the OOM Killer?

OOM is short for Out of Memory. When the Linux system runs out of memory, the kernel activates the OOM Killer mechanism, selectively killing some processes to free up memory and ensure the system doesn't die completely.

The processes that get killed are usually the ones using a lot of memory, such as MySQL, Java applications, Node.js services, and so on. That's why your database or website service suddenly goes down.

Confirm Whether OOM Has Occurred

Check the system logs

# 搜索 OOM 相关日志
dmesg | grep -i "out of memory"
dmesg | grep -i "oom"

# 或者查看 syslog
grep -i "oom" /var/log/syslog
grep -i "oom" /var/log/messages

If you see output similar to this, it means OOM has indeed occurred:

Out of memory: Killed process 1234 (mysqld) total-vm:2048000kB, anon-rss:1024000kB

Check current memory usage

# 查看内存概况
free -h

# 输出示例：
#               total        used        free      shared  buff/cache   available
# Mem:          1.9Gi       1.5Gi       128Mi        64Mi       300Mi       200Mi
# Swap:            0B          0B          0B

Pay close attention to the available column—this is the memory that is truly available. If this value is very small (for example, below 100MB), it means memory really is tight.

Find the biggest memory consumers

# 按内存占用排序显示进程
ps aux --sort=-%mem | head -20

# 或者用 top，按 M 键按内存排序
top

Solutions to the OOM Problem

Solution 1: Add Swap space

If your VPS has no Swap configured, it is strongly recommended that you add it. Swap can provide a buffer when memory runs low; although it is slower, it's better than having processes killed.

# 检查是否有 swap
swapon --show

# 如果没有，创建 2G 的 swap 文件
sudo fallocate -l 2G /swapfile
sudo chmod 600 /swapfile
sudo mkswap /swapfile
sudo swapon /swapfile

# 设置开机自动挂载
echo '/swapfile none swap sw 0 0' | sudo tee -a /etc/fstab

Recommended Swap size:

1GB memory: set Swap to 1-2GB
2GB memory: set Swap to 2GB
4GB or more: setting Swap to 2-4GB is enough

Solution 2: Optimize MySQL memory usage

MySQL is a major memory consumer, and VPSs with small amounts of memory must be tuned. Edit the configuration file:

sudo nano /etc/mysql/mysql.conf.d/mysqld.cnf

Recommended configuration for a 1GB memory VPS:

[mysqld]
# 关键参数
innodb_buffer_pool_size = 128M
innodb_log_buffer_size = 8M
key_buffer_size = 16M

# 连接相关
max_connections = 50
thread_cache_size = 8

# 临时表
tmp_table_size = 16M
max_heap_table_size = 16M

# 查询缓存（MySQL 8.0 已移除）
query_cache_size = 0
query_cache_type = 0

Restart MySQL after making changes:

sudo systemctl restart mysql

Solution 3: Optimize PHP-FPM

If you run WordPress or other PHP applications, the number of PHP-FPM processes also needs to be controlled.

Edit the PHP-FPM configuration:

# 路径可能不同，根据你的 PHP 版本调整
sudo nano /etc/php/8.1/fpm/pool.d/www.conf

Recommended configuration for 1GB memory:

pm = dynamic
pm.max_children = 5
pm.start_servers = 2
pm.min_spare_servers = 1
pm.max_spare_servers = 3
pm.max_requests = 500

Solution 4: Limit the memory usage of a single process

Using systemd, you can limit the maximum memory usage of a service:

# 编辑服务配置
sudo systemctl edit mysql

Add the memory limit:

[Service]
MemoryMax=512M
MemoryHigh=400M

This way MySQL can use at most 512MB of memory, and when it exceeds 400MB the system will try to reclaim memory.

Solution 5: Adjust OOM Killer priority

You can lower the priority of important processes being killed:

# 查看进程的 oom_score（分数越高越容易被杀）
cat /proc/$(pidof mysqld)/oom_score

# 调整 oom_score_adj（范围 -1000 到 1000，-1000 表示永不被杀）
echo -500 > /proc/$(pidof mysqld)/oom_score_adj

To make it permanent, add it to the systemd service file:

[Service]
OOMScoreAdjust=-500

Practical Monitoring Script

Write a simple memory monitoring script that sends an alert when memory is too low:

#!/bin/bash
# /usr/local/bin/memory-monitor.sh

THRESHOLD=100  # 可用内存低于 100MB 时告警

available=$(free -m | awk '/Mem:/ {print $7}')

if [ $available -lt $THRESHOLD ]; then
    echo "警告：可用内存仅剩 ${available}MB" | mail -s "VPS 内存告警" [email protected]
fi

Run it periodically with cron:

# 每 5 分钟检查一次
*/5 * * * * /usr/local/bin/memory-monitor.sh

When Should You Upgrade Your Configuration?

If you have already:

Added Swap
Optimized the memory configuration of each service
Disabled unnecessary services

but memory is still frequently insufficient, then it's time to consider upgrading your VPS configuration. Generally speaking:

A single small site: 1GB of memory is enough
WordPress + a few small services: 2GB is more comfortable
Multiple sites or Java applications: start at 4GB

Summary

The approach to handling OOM problems:

Confirm first - Check the logs to confirm whether OOM has actually occurred
Find the cause - Identify the processes using a lot of memory
Add a buffer - If there's no Swap, add it first
Optimize - Adjust the configuration of services such as MySQL and PHP-FPM
Set limits - If necessary, use systemd to limit process memory
Set up monitoring - Configure alerts so problems are found and handled early

Running services on a small-memory VPS requires careful budgeting, but as long as it's configured properly, 1GB of memory is completely fine for running a personal blog or small project.

# 查看内存概况 free -h # 输出示例： # total used free shared buff/cache available # Mem: 1.9Gi 1.5Gi 128Mi 64Mi 300Mi 200Mi # Swap: 0B 0B 0B

# 检查是否有 swap swapon --show # 如果没有，创建 2G 的 swap 文件 sudo fallocate -l 2G /swapfile sudo chmod 600 /swapfile sudo mkswap /swapfile sudo swapon /swapfile # 设置开机自动挂载 echo '/swapfile none swap sw 0 0' | sudo tee -a /etc/fstab

[mysqld] # 关键参数 innodb_buffer_pool_size = 128M innodb_log_buffer_size = 8M key_buffer_size = 16M # 连接相关 max_connections = 50 thread_cache_size = 8 # 临时表 tmp_table_size = 16M max_heap_table_size = 16M # 查询缓存（MySQL 8.0 已移除） query_cache_size = 0 query_cache_type = 0

# 查看进程的 oom_score（分数越高越容易被杀） cat /proc/$(pidof mysqld)/oom_score # 调整 oom_score_adj（范围 -1000 到 1000，-1000 表示永不被杀） echo -500 > /proc/$(pidof mysqld)/oom_score_adj

#!/bin/bash # /usr/local/bin/memory-monitor.sh THRESHOLD=100 # 可用内存低于 100MB 时告警 available=$(free -m | awk '/Mem:/ {print $7}') if [ $available -lt $THRESHOLD ]; then echo "警告：可用内存仅剩 ${available}MB" | mail -s "VPS 内存告警" [email protected] fi

VPS Out of Memory? OOM Problem Solutions

What Is the OOM Killer?

Confirm Whether OOM Has Occurred

Check the system logs

Check current memory usage

Find the biggest memory consumers

Solutions to the OOM Problem

Solution 1: Add Swap space

Solution 2: Optimize MySQL memory usage

Solution 3: Optimize PHP-FPM

Solution 4: Limit the memory usage of a single process

Solution 5: Adjust OOM Killer priority

Practical Monitoring Script

When Should You Upgrade Your Configuration?

Summary

Loading…

VPS Out of Memory? OOM Problem Solutions

What Is the OOM Killer?

Confirm Whether OOM Has Occurred

Check the system logs

Check current memory usage

Find the biggest memory consumers

Solutions to the OOM Problem

Solution 1: Add Swap space

Solution 2: Optimize MySQL memory usage

Solution 3: Optimize PHP-FPM

Solution 4: Limit the memory usage of a single process

Solution 5: Adjust OOM Killer priority

Practical Monitoring Script

When Should You Upgrade Your Configuration?

Summary