How to Check Linux for Errors: Logs, Disk, Memory, and Network

The system behaves strangely — hangs once a day, the disk is slower than before, a program crashes for no visible reason. Before reinstalling everything in sight — there are tools that show exactly where the problem is.

System Logs: What Happened and When

dmesg — the kernel journal. Hardware messages, drivers, filesystem events. With readable timestamps:

dmesg -T

Errors and warnings only, without informational noise:

dmesg -T --level=err,warn

Watch new messages in real time — useful during diagnostics:

dmesg -Tw

journalctl — the systemd journal. Shows logs for the whole system and individual services.

Errors only from the current boot:

journalctl -p err -b

Errors from the previous boot — useful if the system crashed and rebooted:

journalctl -p err -b -1

Errors from a specific service in the last hour:

journalctl -u nginx --since "1 hour ago" -p err

List all boots — shows when restarts occurred:

journalctl --list-boots

Hard Drive and SSD: Catching Problems Before Failure

Drives do not die suddenly — they warn first. S.M.A.R.T. tracks error counts, reallocated sectors, temperature. Install the tool:

sudo apt install smartmontools

Full health report for a drive:

sudo smartctl -a /dev/sda

List available drives:

lsblk

Key fields to check in smartctl output: Reallocated_Sector_Ct — any non-zero value signals degradation. Current_Pending_Sector — unstable sectors not yet reallocated. Offline_Uncorrectable — sectors that cannot be read. SMART overall-health self-assessment: PASSED/FAILED — the final verdict.

Run a quick built-in test (takes 1-2 minutes):

sudo smartctl -t short /dev/sda

Check the result a few minutes later:

sudo smartctl -l selftest /dev/sda

Filesystem: Check and Repair

fsck checks the filesystem and fixes errors. Important: run only on an unmounted partition.

Force a root filesystem check on next boot:

sudo touch /forcefsck

Check a partition manually (unmount first):

sudo umount /dev/sda1
sudo fsck -f /dev/sda1

The -f flag forces the check even if the filesystem is marked clean. -y automatically fixes all found errors without prompting:

sudo fsck -fy /dev/sda1

RAM: Finding Bad Cells

Memory errors cause the strangest bugs: random program crashes, file corruption, kernel panics with no apparent reason.

memtest86 — a bootable tool that runs before the OS starts. Install to GRUB:

sudo apt install memtest86+
sudo update-grub

Reboot, select memtest86+ from the GRUB menu. A full test takes several hours — best left overnight.

stress-ng — load testing from within the running system:

sudo apt install stress-ng
sudo stress-ng --cpu 4 --io 2 --vm 1 --vm-bytes 512M --timeout 60s --metrics-brief

If the system hangs or crashes during the test — there is a hardware problem. --metrics-brief shows operations per second for each worker.

Network: Drops, Packet Loss, Route

Check basic connectivity with packet loss statistics:

ping -c 20 8.8.8.8

In the statistics line: 0% packet loss — all good. Any percentage of loss — problem somewhere on the route.

Route to host with per-hop latency:

traceroute 8.8.8.8

mtr combines ping and traceroute, updates in real time:

sudo apt install mtr
mtr 8.8.8.8

Errors on network interfaces — corrupted packets, collisions:

ip -s link

If errors and dropped counters are non-zero — problem at the network interface or cable level.

Quick Reference