Linux Problems: Symptoms, Diagnoses, and Fixes

When something on your server is wrong, this is where you start — by symptom, not by guess.

How to Read This Index

When your server has a problem, you usually know the symptom but not the cause. This index is grouped by what you noticed first — the box is slow, the disk is full, a service died, you can't SSH in. Pick the section that matches your symptom and follow the link into the walkthrough. Each problem page diagnoses several possible causes and shows the fix for each one. If you already know exactly which problem you're chasing, jump to the full alphabetical list at the bottom.

1. The Server Is Slow

The classic "something feels wrong" call. The cause is almost always one of these four — load, CPU, memory pressure, or the disk being slow — and they often cascade into each other.

high load — load average way above your core count, and what's actually behind it
high memory usage — RAM is filling up and you're inching toward trouble
high I/O wait — the CPU is waiting on disk, and nothing else can move
thrashing — the box is spending more time paging than working
swapping — memory has overflowed to disk and everything slows to a crawl
runaway process — one process is eating the whole machine
poor performance — the umbrella walkthrough when you don't know which of the above it is
bad performance — the sibling page focused on app-level latency

2. The Disk Is Full or Failing

The single most common production outage. Either the disk filled up, the underlying device is dying, or the filesystem itself has gone sideways.

disk full — out of space, and the inode trap that catches everyone once
disk failing — SMART is screaming; act before you lose data
disk error — reads or writes are returning errors and the kernel log knows why
SSD worn out — the drive has used up its write life; old, not broken — plan the swap
NVMe spare exhausted — the reserve is gone and the drive is about to go read-only; replace now
disk cable errors — rising CRC counts mean a bad cable, not a dying disk
SMART unavailable — can't read the drive's health; here's how to get through
RAID degraded — a drive dropped out of the array; redundancy is gone until you replace it
RAID rebuilding — the array is resyncing onto a fresh disk; what to expect while it runs
file system corruption — the filesystem itself is damaged, and fsck is in your future
data corruption — files, databases, or whole filesystems are returning the wrong bytes

3. Out of Memory, Swap, and Bad RAM

When the kernel runs out of RAM, the OOM killer starts picking processes to sacrifice — or memory pressure spills into swap, or the RAM itself is faulty. This is what you see, and how you get ahead of it.

out of memory — the kernel ran out and started killing things
high memory usage — RAM is filling up; the truth is in MemAvailable, not "used"
memory leak — a process that just keeps eating, forever
swap full — swap is exhausted; with RAM full too, OOM is imminent
swapping — the warning sign before OOM hits
swap thrashing — paging has taken over and nothing useful happens
swappiness misconfigured — the kernel swaps hot memory too eagerly for your workload
memory errors (ECC) — a DIMM is throwing correctable/uncorrectable errors; replace it
bad RAM — the kernel found genuinely faulty memory and poisoned the pages

4. A Service or Process Is Broken

Not the box — just one program. It crashed, it's stuck, it's a zombie, or its code has a bug you finally have to chase.

software crash — a service or daemon died; logs, exit codes, and restart strategy
system crash — the kernel itself went down
segfault — a process touched memory it didn't own, and the kernel killed it
buffer overflow — the classic memory-safety bug, and why it still matters
zombie process — dead but not reaped; what it means and when to worry
device busy — you can't unmount or release something because a process is still holding it
file not found — the file the app wanted isn't where it expected
software bug — the umbrella diagnosis when nothing else fits
configuration error — the code is fine; you mistyped something in a config file
incorrect configuration — the broader walkthrough for misconfigured services

5. The Box Is Unreachable

You can't get in. SSH won't connect, the kernel may have panicked, the network is down, or the firewall is in your way.

system failure — the box stopped responding entirely
kernel panic — the kernel hit something fatal and halted
kernel issue — non-fatal kernel weirdness, modules, parameters
boot failure — it won't come back up; bootloader, initramfs, fstab
SSH issue — keys, configs, sshd flags — when the door won't open
network failure — interfaces, routes, link state; the whole network stack
firewall issue — your rules are blocking what you wanted to allow
DNS issue — names aren't resolving, and everything that depends on them is broken
NTP issue — time sync is broken and TLS is about to start failing
incorrect time — the clock drifted and now auth and certs misbehave

6. Hardware Trouble

When the silicon underneath has stopped cooperating — disks, RAM, power, the board itself.

hardware failure — the umbrella for physical-layer problems
disk failing — SSDs and HDDs both warn before they die; listen
power issue — outages, brownouts, PSU trouble

7. Security

When something — or someone — got in, or could.

security breach — they're already inside; here's how you find them and what to do
data breach — data has left the building; obligations, forensics, notifications
security vulnerability — a known hole you need to close before it's used
security issue — the umbrella walkthrough for weaker-than-it-should-be
malware infection — something malicious is running; spot it and remove it
denial of service — the box is being drowned in traffic on purpose
permission issue — the wrong user can read/write something they shouldn't, or the right user can't

8. Application-Level Problems

Higher up the stack — your database, your email server, your app — when the OS is fine but the workload is not.

database issue — slow queries, replication lag, deadlocks, corruption
email issue — sending, receiving, deliverability, queues backed up
virtualization issue — guests misbehaving, hypervisor quirks, noisy neighbors

Full Alphabetical List

Every problem page in the knowledge base. If you already know the name, jump straight to it.

bad performance — the app-latency-focused performance walkthrough
bad RAM — the kernel found genuinely faulty memory and poisoned the pages
boot failure — the server won't come back up after a reboot
buffer overflow — the classic memory-safety bug and what it still costs us
configuration error — a misconfigured service refusing to start or behaving oddly
data breach — sensitive data was accessed or exposed without authorization
data corruption — files, databases, or filesystems returning the wrong bytes
database issue — slow queries, replication lag, deadlocks, corruption
denial of service — the box is being drowned in traffic on purpose
device busy — a resource is still held open and you can't release it
disk cable errors — rising CRC counts point at the cable, not the disk
disk error — read/write errors surfacing in the kernel ring buffer
disk failing — SMART is warning; act before you lose data
disk full — out of space, and the inode trap that catches everyone once
DNS issue — names aren't resolving and everything downstream breaks
email issue — sending, receiving, deliverability, queue trouble
file not found — the file the application expected isn't there
file system corruption — the filesystem itself is damaged
firewall issue — your rules are blocking what you wanted to allow
hardware failure — the umbrella for physical-layer problems
high I/O wait — the CPU is waiting on disk and nothing else can move
high load — load average way above core count, and what's behind it
high memory usage — RAM is filling up and you're approaching trouble
incorrect configuration — broader walkthrough for misconfigured services
incorrect time — the clock drifted and now auth and certs misbehave
kernel issue — non-fatal kernel weirdness, modules, parameters
kernel panic — the kernel hit something fatal and halted
malware infection — something malicious is running on the box
memory errors (ECC) — a DIMM is throwing correctable/uncorrectable errors; replace it
memory leak — a process that just keeps eating RAM forever
network failure — interfaces, routes, link state; the whole network stack
NTP issue — time sync is broken and TLS is about to start failing
NVMe spare exhausted — the reserve ran dry; the drive is a real defect, replace now
out of memory — the kernel ran out and started killing things
permission issue — the wrong access or denied access where you didn't expect
poor performance — the umbrella walkthrough for "it's slow"
power issue — outages, brownouts, PSU trouble
RAID degraded — a drive dropped out of the array; redundancy is gone
RAID rebuilding — the array is resyncing onto a fresh disk
runaway process — one process is eating the whole machine
security breach — they're already inside; find them and respond
security issue — the umbrella walkthrough for weaker-than-it-should-be
security vulnerability — a known hole you need to close
segfault — a process touched memory it didn't own
SMART unavailable — can't read the drive's health log, and how to get through
software bug — the umbrella diagnosis when nothing else fits
software crash — a service or daemon died; logs, exit codes, restarts
SSD worn out — the drive used up its write life; old, not broken — plan the swap
SSH issue — keys, configs, sshd flags — when the door won't open
swap full — swap is exhausted; with RAM full too, OOM is imminent
swap thrashing — the box pages constantly and does no real work
swappiness misconfigured — the kernel swaps hot memory too eagerly for the workload
swapping — memory has overflowed to disk and everything slows to a crawl
system crash — the kernel itself went down
system failure — the box stopped responding entirely
virtualization issue — guests misbehaving, hypervisor quirks
zombie process — dead but not reaped; what it means and when to worry

Watching the right symptoms early prevents most of these from ever happening.

CleverUptime watches the underlying signals — disk usage trending up, load creeping, memory pressure rising, services flapping, errors landing in the kernel ring buffer — long before they turn into the symptoms above, and tells you in plain language what's wrong so you reach the right diagnosis fast.

Want to see your own server's health right now? One command, no signup, no install.

Check your server →