When deploying Linux systems on industrial CompactFlash (CF) cards that must operate unattended for extended periods, filesystem reliability becomes paramount. The primary concerns are:

• Minimizing write operations to prolong flash memory lifespan
• Ensuring filesystem consistency after power failures
• Reducing or eliminating the need for fsck on boot

For this specific use case, we recommend these mount options in /etc/fstab:

UUID=xxxx-xxxx  /  ext4  defaults,noatime,nodiratime,data=journal,commit=60,barrier=1  0  1

Breaking down the critical options:

• data=journal: Provides full data and metadata journaling for maximum reliability
• commit=60: Increases commit interval to reduce writes (tune based on your risk tolerance)
• noatime/nodiratime: Eliminates unnecessary access time updates
• barrier=1: Ensures proper write ordering (critical for flash storage)

When initially creating the filesystem:

mkfs.ext4 -O ^has_journal,^extent,^flex_bg -E lazy_itable_init=0,lazy_journal_init=0 /dev/sdX
tune2fs -o journal_data /dev/sdX
tune2fs -c 0 -i 0 /dev/sdX

These parameters:

• Explicitly enable journaling after creation for better control
• Disable extents which can complicate recovery
• Remove unnecessary filesystem checks

For additional protection:

echo 5 > /proc/sys/vm/dirty_background_ratio
echo 10 > /proc/sys/vm/dirty_ratio
echo 6000 > /proc/sys/vm/dirty_expire_centisecs
echo 300 > /proc/sys/vm/dirty_writeback_centisecs

These kernel parameters help:

• Reduce dirty page thresholds
• Shorten writeback intervals
• Minimize potential data loss windows

Implement a simple monitoring script to check filesystem health:

#!/bin/bash
# Check filesystem errors
ERR_COUNT=$(dmesg | grep -i "ext4.*error" | wc -l)
if [ $ERR_COUNT -gt 0 ]; then
    logger -t fscheck "EXT4 errors detected in kernel log"
fi

# Force periodic sync if needed
sync

Schedule this to run periodically via cron to catch potential issues early.

For maximum reliability, consider making the root filesystem read-only:

mount -o remount,ro /

Then configure a separate small partition with these fstab options for writable data:

UUID=yyyy-yyyy  /data  ext4  defaults,noatime,data=journal,commit=60,barrier=1  0  2

This approach:

• Completely protects the root filesystem
• Confines writes to a designated area
• Minimizes potential corruption points

---

When configuring ext4 for unattended industrial systems using CompactFlash storage, we must prioritize these technical considerations:

• Data integrity over performance
• Power loss resilience as a primary design goal
• Minimal maintenance (ideally zero fsck requirements)

These mount options in /etc/fstab provide the best balance for embedded systems:

UUID=xxxx-xxxx / ext4 errors=remount-ro,data=journal,nodelalloc,noauto_da_alloc 0 1

Key options explained:

• data=journal - Full journaling including file data (slowest but safest)
• nodelalloc - Disables delayed allocation to prevent metadata corruption
• noauto_da_alloc - Prevents unsafe optimizations that could cause corruption

Add these to your boot parameters (/etc/default/grub):

GRUB_CMDLINE_LINUX="rootflags=data=journal elevator=noop"

When initially creating the filesystem (or when reformatting):

mkfs.ext4 -O ^has_journal,^extent,^uninit_bg -E lazy_itable_init=0 /dev/sdX

Then re-enable journaling separately:

tune2fs -o journal_data /dev/sdX

Even with these precautions, schedule monthly maintenance:

#!/bin/bash
# Force filesystem sync and journal flush
sync
echo 3 > /proc/sys/vm/drop_caches

Here's a complete example for an embedded industrial controller:

# /etc/fstab
UUID=5e4c4d4e-1234-5678-90ab-cdef01234567 / ext4 defaults,errors=remount-ro,data=journal,nodelalloc,noauto_da_alloc,commit=60 0 1

# Kernel parameters in /etc/default/grub
GRUB_CMDLINE_LINUX_DEFAULT="quiet rootflags=data=journal elevator=noop"

Implement this simple monitoring script:

#!/bin/bash
# Check filesystem health
HEALTH=$(dmesg | grep "EXT4-fs error")
if [ -n "$HEALTH" ]; then
    logger -t ext4mon "Filesystem errors detected: $HEALTH"
fi