When dealing with secondary storage that needs to remain mounted but inactive, we face three technical challenges: process monitoring, kernel-level power management, and disk sleep configuration. Let's break this down systematically.

To identify what's waking your disk, Linux provides several powerful tools:

# Method 1: Using fatrace (File Activity Trace)
sudo apt install fatrace
sudo fatrace | grep "/mount/point/of/your/disk"

# Method 2: Using inotifywait
sudo apt install inotify-tools
inotifywait -m -r /mount/point | grep --line-buffered "OPEN"

# Method 3: Using systemtap (advanced)
stap -e 'probe vfs.read,vfs.write {
    if (d_name($path) == "your_disk_mount_point")
        printf("%s %s\\n", execname(), probefunc())
}'

Modern Linux kernels (4.15+) include several parameters for fine-tuning disk sleep behavior. These are particularly important for preventing unnecessary wakeups:

# Check current disk power management settings
hdparm -B /dev/sdX
hdparm -S /dev/sdX

# Recommended settings for our use case:
sudo hdparm -B 127 /dev/sdX  # Balanced power mode
sudo hdparm -S 241 /dev/sdX  # 30-minute standby timeout

The key kernel parameters to adjust in /etc/sysctl.conf:

vm.laptop_mode = 5
vm.dirty_writeback_centisecs = 6000
vm.dirty_expire_centisecs = 6000

For persistent configuration that survives reboots, we can create a udev rule:

# /etc/udev/rules.d/99-hdd-power.rules
ACTION=="add", SUBSYSTEM=="block", KERNEL=="sd[b-z]", RUN+="/usr/bin/hdparm -B 127 -S 241 /dev/%k"

Certain filesystem features can prevent proper disk sleeping:

# For ext4, disable unnecessary access time updates
sudo tune2fs -o nobarrier,noatime,nodiratime /dev/sdX1

# If using NTFS (via ntfs-3g), add these mount options:
rw,noatime,nodiratime,users,umask=000,uid=1000,gid=1000

Create a service to manage disk state more intelligently:

# /etc/systemd/system/hdd-sleep.service
[Unit]
Description=HDD Sleep Manager

[Service]
Type=oneshot
ExecStart=/usr/bin/hdparm -y /dev/sdX
ExecStop=/usr/bin/hdparm -S 241 /dev/sdX

[Install]
WantedBy=multi-user.target

To confirm your disk is sleeping properly:

# Check disk power state
sudo smartctl -n standby /dev/sdX

# Continuous monitoring
watch -n 60 'hdparm -C /dev/sdX | grep -i state'

Remember that some applications (like backup tools or desktop indexers) may still cause periodic wakeups. The monitoring tools mentioned earlier will help identify these offenders.

---

When dealing with secondary storage devices that need to remain mounted but inactive most of the time, we face two primary technical challenges: preventing unnecessary wake-ups and identifying what causes them. This requires a combination of kernel-level power management and detailed access monitoring.

To track which processes access your HDD, we'll use Linux's audit subsystem. First, install the necessary tools:

sudo apt install auditd
sudo systemctl enable --now auditd

Then create a rule to monitor your device (replace /dev/sdX with your actual device):

sudo auditctl -w /dev/sdX -p rwa -k hdd_access

To view the logs:

sudo ausearch -k hdd_access | aureport -f -i

The key parameters controlling disk spindown are in /sys/block/sdX/device/:

echo 10 > /sys/block/sdX/device/power/autosuspend_delay_ms
echo auto > /sys/block/sdX/device/power/control

For more aggressive power management, add these kernel boot parameters:

GRUB_CMDLINE_LINUX_DEFAULT="quiet splash usbcore.autosuspend=-1 scsi_mod.use_blk_mq=1"

hdparm provides direct control over drive power states:

sudo hdparm -S 60 /dev/sdX  # 5-minute timeout (60 * 5 seconds)
sudo hdparm -B 127 /dev/sdX  # Balance between performance and power saving

Certain filesystem operations can trigger wake-ups. For ext4:

tune2fs -o journal_data_writeback /dev/sdX1
tune2fs -E mount_opts=auto_da_alloc /dev/sdX1

Create a custom systemd unit to mount the drive only when needed:

[Unit]
Description=Mount secondary HDD on demand
Requires=network-online.target
After=network-online.target

[Service]
Type=oneshot
RemainAfterExit=yes
ExecStart=/usr/bin/mount /mnt/secondary
ExecStop=/usr/bin/umount /mnt/secondary

[Install]
WantedBy=multi-user.target

Common culprits and solutions:

# Disable udisks2 polling
sudo systemctl mask udisks2.service

# Prevent cron jobs from scanning
sudo touch /etc/cron.d/noscan
sudo chattr +i /etc/cron.d/noscan

Combine audit logs with power state monitoring:

#!/bin/bash
DEVICE="/dev/sdX"
LOG_FILE="/var/log/hdd_monitor.log"

while true; do
  STATE=$(cat /sys/block/${DEVICE##*/}/device/power/runtime_status)
  if [[ "$STATE" != "suspended" ]]; then
    TIMESTAMP=$(date +"%Y-%m-%d %T")
    PROCESSES=$(sudo lsof $DEVICE)
    echo "[$TIMESTAMP] Device active" >> $LOG_FILE
    echo "$PROCESSES" >> $LOG_FILE
  fi
  sleep 5
done