How to Check Uptime in Linux: uptime, w, and Beyond

Uptime tells you how long the system has been running — and the load average tells you how hard it's been working. Both are packed into a single line.

---

TL;DR

uptime              # uptime + load average
w                   # uptime + logged in users + their activity
last reboot         # when the system was last restarted
cat /proc/uptime    # uptime in seconds (for scripting)

---

uptime Command

uptime

 09:15:42 up 14 days,  2:31,  3 users,  load average: 1.24, 0.87, 0.63

Breaking it down:

Part	Meaning
09:15:42	Current time
up 14 days, 2:31	System running for 14 days, 2 hours, 31 minutes
3 users	3 active sessions (SSH + console)
load average: 1.24, 0.87, 0.63	1, 5, 15-minute averages

Uptime with one-liner format

uptime -p          # human-readable duration only
# up 2 weeks, 14 hours, 31 minutes

uptime -s          # boot time as timestamp
# 2026-04-08 06:44:11

---

Load Average Explained

load average: 1.24, 0.87, 0.63
              1min  5min  15min

Load average = average number of processes either running or waiting to run (including I/O-waiting processes).

How to read it:

nproc    # number of CPU cores
# 4 cores

# load 1.24 on 4 cores = 31% utilization → healthy
# load 4.00 on 4 cores = 100% utilization → at capacity
# load 6.50 on 4 cores = 163% utilization → overloaded

Trend reading:

• 1.24, 0.87, 0.63 — load rising (1min > 15min) → system getting busier
• 0.50, 1.20, 2.10 — load falling (1min < 15min) → recovering from spike
• 1.24, 1.25, 1.23 — load stable

---

w: Uptime + Active Users

w

 09:15:42 up 14 days,  2:31,  3 users,  load average: 1.24, 0.87, 0.63
USER     TTY      FROM             LOGIN@   IDLE JCPU   PCPU WHAT
damon    pts/0    10.0.1.42        09:00    0.00s  0.10s  0.01s w
ops      pts/1    192.168.1.100    08:30    5:23   0.20s  0.00s bash
deploy   pts/2    10.0.1.50        09:10    0.00s  2.10s  2.09s npm run build

Useful for seeing who's connected and what they're running — especially during an incident where you want to know if someone is actively working on the server.

Columns: IDLE = how long since last input, PCPU = current CPU usage of the command.

---

Find When the System Last Rebooted

# Last reboot event
last reboot | head -5

# Output:
# reboot   system boot  5.15.0-91-gene Mon Apr  8 06:44   still running
# reboot   system boot  5.15.0-91-gene Sun Mar 31 14:22 - 06:44 (9+16:21)

# Using who
who -b
# system boot  2026-04-08 06:44

# Using uptime -s
uptime -s
# 2026-04-08 06:44:11

---

/proc/uptime for Scripting

cat /proc/uptime
# 1234567.89 4987654.21
# First number = uptime in seconds
# Second number = idle time in seconds (across all CPUs)

Convert to human-readable in a script:

seconds=$(awk '{print int($1)}' /proc/uptime)
days=$((seconds / 86400))
hours=$(( (seconds % 86400) / 3600 ))
minutes=$(( (seconds % 3600) / 60 ))
echo "Uptime: ${days}d ${hours}h ${minutes}m"

---

Real Examples

Check if a server was recently rebooted (monitoring script)

#!/bin/bash
# Alert if server rebooted in the last 10 minutes
uptime_seconds=$(awk '{print int($1)}' /proc/uptime)
if [ "$uptime_seconds" -lt 600 ]; then
  echo "ALERT: Server rebooted $(($uptime_seconds / 60)) minutes ago"
  last reboot | head -2
fi

Check uptime across multiple servers

for host in web01 web02 db01; do
  echo -n "$host: "
  ssh $host 'uptime -p'
done

Correlate uptime with load average trend

# Is load average trending up over the past 15 minutes?
uptime | awk '{
  split($NF, a, ",");
  load1=a[1]; load15=a[3];
  if (load1 > load15 * 1.2)
    print "Load rising: 1min=" load1 " 15min=" load15
  else
    print "Load stable: 1min=" load1 " 15min=" load15
}'

---

Common Mistakes

Mistake 1: High load average always means CPU problem Load average includes I/O-waiting processes. A server with 20 processes stuck on a hung NFS mount shows load 20 with CPU at 5%. Check %wa in top.

Mistake 2: Not accounting for CPU count when reading load Load average of 4.0 is:

• Critical on a 2-core server (200% utilization)
• Healthy on an 8-core server (50% utilization)

Always check nproc alongside load average.

Mistake 3: Using uptime to determine if a service restarted uptime shows OS uptime, not service uptime. A service can crash and restart while the OS keeps running.

# Service uptime, not OS uptime
systemctl status nginx | grep "Active:"
# Active: active (running) since Mon 2026-04-20 14:22 UTC; 1h 30min ago

---

Pro Tips

# Compact uptime summary
uptime | awk '{printf "Up: %s %s, Load: %s %s %s\n", $3, $4, $(NF-2), $(NF-1), $NF}'

# Monitor load average with timestamps
while true; do
  echo "$(date '+%H:%M:%S') - $(uptime | awk -F'load average:' '{print $2}')"
  sleep 10
done

# Find the last 5 reboots
last reboot | head -5

# Show uptime in seconds (exact, for monitoring)
awk '{print int($1)}' /proc/uptime

---

Conclusion

uptime in one line gives you OS uptime and load average trend. The load average only means something relative to your CPU count — always check nproc. A rising trend (1min > 15min) means the system is getting busier; a falling trend means it's recovering. For scripting, /proc/uptime gives you seconds directly without parsing.

---

Related: How to Check CPU Usage in Linux — diagnose high load average with CPU investigation tools. Linux Check Running Services with systemctl — check what's been running since the last boot.