---
title: dmesg — check kernel messages
date: 2026-10-18
time: 12:00
model: admin
category: programs
author_type: human
summary: Command that checks hardware, driver, and OOM messages left by the kernel. Essential for diagnosing unexplained reboots.
tags: program, server log recording and analysis, dmesg
---

**💬 "dmesg command"**

dmesg displays **kernel ring buffer** messages. Disk errors, network link issues, and OOM kills are recorded here.

Because it is separate from user-space application logs, you should review it together when troubleshooting unexplained problems.

- **Concept**: Shows kernel messages in chronological order.
- **Left**: Kernel time in the format [  12.345678] (seconds elapsed since boot).
- **Contents**: Drivers, disk I/O errors, OOM killer, link status, and more.
- **-T**: Converts to human-readable actual time.
## Commonly used options

| Option | Description |
| --- | --- |
| -T | Convert timestamps to actual time |
| -l err,warn | Show only specific levels |
| -k | Show only kernel messages |
| -w / --follow | Follow new messages in real time |

## 3 essential command combinations

- `dmesg -T | tail -50` — Check recent kernel messages
- `dmesg -T | grep -i -E 'oom|killed process'` — Check whether OOM kills occurred
- `dmesg -T | grep -i -E 'error|fail|i/o'` — Find hardware and I/O errors
**💬 "Show me an example of safely reverting this command"**

Here are situations commonly encountered on live production servers, organized in order.

### Step 1 · Check recent messages

View recent logs related to boot and devices.

```
dmesg -T | tail -50
```

### Step 2 · Check for OOM

Check whether a process was killed due to insufficient memory.

```
dmesg -T | grep -i -E 'oom|killed process'
```

### Step 3 · Check disk errors

Check for I/O errors.

```
dmesg -T | grep -i -E 'i/o error|medium error|ata'
```

## How to use this in practice

### 1. Process suddenly dies

**Situation** — An app process terminates without warning.

```
dmesg -T | grep -i -E 'oom|killed process'
```

**Result** — If you see the message Out of memory: Killed process, the OOM killer terminated the process.

**Point** — Also check memory usage, and adjust swap or memory limits if needed.

### 2. Unexplained reboot

**Situation** — There is a record that the server rebooted in the early morning.

```
dmesg -T | head -30
```

**Result** — In the early boot messages, check hardware initialization and traces of the previous shutdown.

**Point** — When used together with journalctl --list-boots, you can identify the exact reboot time.

## Important notes

- The ring buffer has a limited size, so old messages disappear. Use journal/rsyslog for permanent retention.
- Kernel time alone makes it hard to know the actual time, so use -T.
- Some distributions restrict dmesg access with kernel.dmesg_restrict.
## Actual output

This is the result of checking for OOM kills and disk I/O errors.

![dmesg command output: kernel OOM and I/O errors](https://games.aidebatehub.com/assets/dmesg-demo.png)

Live production server

---
Source (HTML): https://games.aidebatehub.com/programs/linux/logs/dmesg.html
This file is an auto-generated Markdown twin of the DevHub document.