On my Red Hat Enterprise Linux 8 server with 8GB RAM, I observed something puzzling:

$ free -h
              total        used        free      shared  buff/cache   available
Mem:          7.7Gi       3.2Gi       487Mi       1.0Gi       4.0Gi       3.2Gi

Despite processes only consuming ~3.2GB, free memory appears alarmingly low at 487MB. Let's decode what's really happening.

Modern Linux systems (including RHEL) employ sophisticated memory management techniques:

• Page Cache: Filesystem data cached in RAM
• Slab Allocator: Kernel object caching
• Buffers: Temporary storage for block I/O

This explains our earlier free output where buff/cache consumed 4GB.

For accurate memory analysis, use these commands:

# Detailed memory breakdown
$ cat /proc/meminfo

# Per-process memory usage
$ ps aux --sort=-%mem | head

# Slab allocation details
$ slabtop -o

# Page cache statistics
$ vmstat -s

The key metrics to watch:

MemFree:      499236 kB
MemAvailable: 3401296 kB
Buffers:       12452 kB
Cached:      3184944 kB

MemAvailable is the critical value - it estimates memory available for new processes without swapping. In this case, we actually have ~3.4GB available despite the low free memory.

For monitoring scripts, use this Python snippet to get accurate available memory:

import re

def get_available_memory():
    with open('/proc/meminfo') as f:
        meminfo = f.read()
    match = re.search(r'MemAvailable:\s+(\d+)\s+kB', meminfo)
    if match:
        return int(match.group(1)) / 1024  # Convert to MB
    return 0

print(f"Available memory: {get_available_memory():.2f} MB")

Real memory pressure indicators:

$ vmstat 1 5
procs -----------memory---------- ---swap--
 r  b   swpd   free   buff  cache
 1  0      0 487236  12452 3184944

Watch for:

• High si (swap in) or so (swap out) values
• Consistently low MemAvailable
• High buff/cache with active swapping

For specialized workloads, consider adjusting:

# Check current settings
$ sysctl vm.swappiness vm.vfs_cache_pressure

# Temporary adjustments
$ sudo sysctl -w vm.swappiness=10
$ sudo sysctl -w vm.vfs_cache_pressure=50

These values control how aggressively Linux reclaims memory from cache versus swapping.

What appears as "missing" memory is actually being efficiently utilized by the kernel. The free command's output is often misinterpreted - focus on available rather than free memory for accurate assessment of your system's memory state.

---

Many Red Hat Linux administrators encounter this puzzling scenario: system monitoring tools report low free memory while process memory consumption doesn't account for all allocated RAM. Here's what's really happening under the hood.

Modern Linux kernels (including Red Hat's) aggressively utilize unused RAM for disk caching and buffers to improve performance. This cached memory appears as "used" in tools like free -m but is actually immediately available when applications need it.

Consider this typical free output:

$ free -m
              total        used        free      shared  buff/cache   available
Mem:           7982        3052         243         125        4686        4251
Swap:          2047           0        2047

• Used: Includes application memory PLUS caches/buffers
• Buff/cache: Disk cache that can be reclaimed instantly
• Available: The true free memory including reclaimable cache

To see memory allocation details, use:

$ cat /proc/meminfo
MemTotal:        8175204 kB
MemFree:          248932 kB
MemAvailable:    4352748 kB
Buffers:          142844 kB
Cached:          3927244 kB
SwapCached:            0 kB
...

Database systems like MySQL/PostgreSQL often exhibit this behavior because:

1. They use large shared memory segments
2. Linux caches frequently accessed database files

Check shared memory with:

$ ipcs -m
------ Shared Memory Segments --------
key        shmid      owner      perms      bytes      nattch     status      
0x0052e2c1 327680     mysql      600        256000000  10

For database servers, consider adjusting these kernel parameters in /etc/sysctl.conf:

vm.swappiness = 10
vm.vfs_cache_pressure = 50
vm.dirty_ratio = 20
vm.dirty_background_ratio = 10

Instead of just checking free memory, use these more accurate commands:

$ vmstat -s
$ cat /proc/meminfo | grep -i available
$ ps aux --sort=-%mem | head

For persistent monitoring, configure tools like:

$ sudo dnf install procps-ng sysstat
$ sar -r 1 3  # Sample memory usage every 1 second, 3 times

Actual memory pressure indicators include:

• Consistently low "Available" value in free
• High swap usage despite available RAM
• OOM killer activating frequently

For advanced analysis, use:

$ slabtop -o
$ pmap -x [pid]
$ numastat -m