How to Check Memory Usage in Linux (free, top, meminfo)

Updated: 11 October 2026 · Applies to: Debian 13, Ubuntu 24.04 and 26.04, AlmaLinux 10, Rocky Linux 9 and Fedora 44 (procps-ng 3.3.17 to 4.0.6, htop 3.3 and 3.4, systemd 259)

To check memory usage in Linux, run free -h and look at the available column: that is the memory programs can still use. Linux fills spare RAM with disk cache (buff/cache) and hands it back as soon as a program needs it, so a small free figure is normal and not a problem. For a live view sorted by memory, run top and press Shift+M.

1. Check memory usage with free

$ free -h
               total        used        free      shared  buff/cache   available
Mem:           3.7Gi       621Mi       842Mi        71Mi       2.6Gi       3.1Gi
Swap:          1.2Gi       6.2Mi       1.2Gi

This is a lab example from our 4 GB test machine. Only 842 MiB is completely unused, but 3.1 GiB is available, because most of the 2.6 GiB cache can be released. What each column means, according to the free manual (procps-ng 4.0.4):

• total: usable RAM (and swap on the second line).
• used: used or unavailable memory, calculated as total minus available.
• free: memory nobody uses at all, not even the cache.
• shared: mostly tmpfs, such as /dev/shm and a RAM-based /tmp.
• buff/cache: kernel buffers, page cache and reclaimable slab. Add -w to see buffers and cache in separate columns.
• available: an estimate of how much memory new programs can get without swapping. Watch this one.

Useful options, all tested on Debian 13:

free -m            # in MiB instead of auto units
free -h -t         # adds a Total line for RAM + swap
free -h -s 2 -c 5  # print every 2 seconds, 5 times
free -h -L         # one line: SwapUse, CachUse, MemUse, MemFree
free -h --si       # powers of 1000 (4.0G) instead of 1024 (3.7Gi)

To get the percentage of memory in use (total minus available, divided by total):

$ free | awk '/Mem:/ {printf "%.1f%% used\n", ($2-$7)/$2*100}'
16.3% used

On minimal AlmaLinux, Rocky Linux and Fedora images, the shell answered free: command not found. Install it with sudo dnf install procps-ng (on Debian and Ubuntu the package is procps).

2. Read the raw numbers in /proc/meminfo

free, top and most monitoring tools read /proc/meminfo. You can read it yourself, which is handy in scripts:

$ grep -E 'MemTotal|MemFree|MemAvailable|^Cached|Buffers|Swap(Total|Free)' /proc/meminfo
MemTotal:        3910520 kB
MemFree:          863008 kB
MemAvailable:    3273592 kB
Buffers:          149344 kB
Cached:          2185880 kB
SwapTotal:       1228776 kB
SwapFree:        1222428 kB

According to the Linux kernel documentation, MemAvailable is "an estimate of how much memory is available for starting new applications, without swapping", calculated from free memory, reclaimable slab, the file cache and the low watermarks. This one-liner prints the used percentage and the available memory from it:

$ awk '/MemTotal/ {t=$2} /MemAvailable/ {a=$2} END {printf "Used: %.1f%%  Available: %.0f MiB\n", (t-a)/t*100, a/1024}' /proc/meminfo
Used: 16.3%  Available: 3197 MiB

3. Watch memory live with top or htop

Run top. The two header lines show the same figures as free (avail Mem is the available column). Press Shift+M to sort processes by %MEM, and q to quit. For a one-off list, for example in a ticket or a script, use batch mode:

$ top -b -n 1 -o %MEM | head -5
top - 12:24:57 up 11 days,  5:55,  0 users,  load average: 1.68, 1.99, 1.18
Tasks:   5 total,   1 running,   3 sleeping,   0 stopped,   1 zombie
%Cpu(s):  0.0 us,  0.0 sy,  0.0 ni,100.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
MiB Mem :   3818.9 total,    842.8 free,    622.0 used,   2707.6 buff/cache
MiB Swap:   1200.0 total,   1193.8 free,      6.2 used.   3196.9 avail Mem

In the process list, RES is the physical memory a process uses and VIRT is everything it has reserved, which is usually much larger and not a reason to worry. More options are in top command usage and examples.

htop shows the same with coloured memory bars. Install it with sudo apt install htop on Debian and Ubuntu, sudo dnf install htop on Fedora, and on AlmaLinux or Rocky Linux enable EPEL first (sudo dnf install epel-release). Press F6 to choose the sort column.

4. See memory trends and swapping with vmstat

$ vmstat -S M 1 3
procs -----------memory---------- ---swap-- -----io---- -system-- -------cpu-------
 r  b   swpd   free   buff  cache   si   so    bi    bo   in   cs us sy id wa st gu
 0  0      6    842    145   2561    0    0    14   196   75    1  1  1 95  0  3  0
 0  0      6    842    145   2561    0    0     0     0   81  154  0  0 100  0  0  0
 0  0      6    842    145   2561    0    0     0 10624  309  701  3  7 81  2  7  0

This prints three reports one second apart, in MiB. The first line is the average since boot; the next lines show the current state. vmstat -S M 5 keeps printing every 5 seconds until you press Ctrl+C. swpd is swap in use; si and so are memory swapped in and out per second. Occasional non-zero values are fine, but if si and so stay above zero the server is short of RAM and will feel slow. vmstat -s -S M prints a summary list instead. See understand more about vmstat.

5. Check the memory usage of a process or service

1. The ten processes using the most memory (RSS is in KiB):

   ps -eo pid,user,%mem,rss,comm --sort=-rss | head -n 11

2. One process by its PID:

   $ ps -o pid,rss,vsz,%mem,cmd -p 5513
       PID   RSS    VSZ %MEM CMD
      5513  1408   2580  0.0 sleep 300
   $ grep -E 'VmRSS|VmHWM|VmSwap' /proc/5513/status
   VmHWM:	    1408 kB
   VmRSS:	    1408 kB
   VmSwap:	       0 kB

   VmHWM is the highest RSS the process has reached and VmSwap is how much of it is in swap. Find the PID with pgrep nginx or pidof nginx.
3. A whole service, including all its child processes (systemd counts the service's control group):

   $ systemctl status cron --no-pager | grep Memory
        Memory: 356K (peak: 1.8M)

   systemd-cgtop -m lists all services sorted by memory.
4. RSS counts shared libraries in full for every process, so adding up RSS overstates the total. On Debian and Ubuntu, smem (package smem) shows PSS, which divides shared memory fairly: smem -t -k.

For more ways to rank processes see find top running processes by memory and CPU usage and list running processes with ps, top, htop and pgrep.

6. Check swap and installed RAM

$ swapon --show
NAME      TYPE      SIZE USED PRIO
/dev/sda2 partition 1.2G 6.2M   -2

No output means the server has no swap; see increase swap space with a swap file. To see the installed memory, sudo lshw -short -C memory printed 4GiB System memory in our test, while dmidecode -t memory needs access to real hardware and failed inside a container. On a VPS, total is the RAM of your plan; Virtualizor also shows it in the panel (view VPS details in Virtualizor).

Ubuntu desktop and containers

• Ubuntu desktop: the commands above work in Terminal. The graphical way is the System Monitor app, Resources tab, which draws a Memory and Swap graph (gnome-system-monitor 49 on Ubuntu 26.04).
• Docker or Podman containers: free inside a container shows the host's memory. In our test a container limited to 256 MB still showed 3.7Gi total. Read the limit and the real usage from the cgroup instead: cat /sys/fs/cgroup/memory.max /sys/fs/cgroup/memory.current (bytes).

When memory runs out

If available memory falls close to zero and swap is full, the kernel's OOM killer stops a process to save the server. Check the kernel log for it:

sudo journalctl -k | grep -i "killed process"

According to the kernel source, the log line reads Out of memory: Killed process followed by the process ID and the program name. See what is the OOM killer. Then reduce the memory the services use, or move to a plan with more RAM. Clearing the cache by hand rarely helps, because Linux already gives cache back when programs need it; when it does make sense, see how to free up Linux memory safely.

Official documentation: free(1) and the /proc filesystem (meminfo).