After configuring ZFS (version 0.6.5) on CentOS 7 and creating a zpool, the system fails to maintain datasets across reboots. The zfs list command returns "no datasets available" and zpool list shows "no pools available" post-reboot.

The zfs-import-cache service appears inactive, despite being loaded:

zfs-import-cache.service - Import ZFS pools by cache file
Loaded: loaded (/usr/lib/systemd/system/zfs-import-cache.service; static)
Active: inactive (dead)

While manual import works using:

zpool import -c /etc/zfs/zpool.cache

And setting the cache file path helps temporarily:

zpool set cachefile=/etc/zfs/zpool.cache PoolName

This solution isn't persistent and requires manual intervention after each reboot.

The /etc/sysconfig/zfs configuration shows standard settings, but key parameters might need adjustment:

# Run zfs mount during system start?
ZFS_MOUNT='yes'

# Run zfs unmount during system stop?
ZFS_UNMOUNT='yes'

Option 1: Enable Cache Service Persistently

First, verify the cache file exists and contains your pool:

ls -l /etc/zfs/zpool.cache
hexdump -C /etc/zfs/zpool.cache

Then enable and start the service:

systemctl enable zfs-import-cache
systemctl start zfs-import-cache

Option 2: Alternative Import Method

Configure the system to import by device ID instead of cache file:

systemctl disable zfs-import-cache
systemctl enable zfs-import.target
systemctl enable zfs-import-scan.service

After implementing changes, verify with:

systemctl status zfs-import*
zpool status
zfs list

Check the boot sequence logs for ZFS-related errors:

journalctl -b | grep -i zfs
dmesg | grep -i zfs

For more detailed debugging, increase ZFS verbosity:

echo 1 > /sys/module/zfs/parameters/zfs_dbgmsg_enable
zpool events

Create custom systemd units for critical datasets:

[Unit]
Description=Mount ZFS dataset
After=zfs-import.target

[Service]
Type=oneshot
ExecStart=/usr/bin/zfs mount pool/dataset
TimeoutSec=0

[Install]
WantedBy=multi-user.target

---

After configuring ZFS 0.6.5 on CentOS 7, I encountered a persistent issue where all datasets and pools become unavailable after system reboot. Running zfs list returns "no datasets available" and zpool list shows "no pools available".

The temporary solution involves manual cache file operations:

# Before reboot
zpool set cachefile=/etc/zfs/zpool.cache PoolName

# After reboot
zpool import -c /etc/zfs/zpool.cache

The zfs-import-cache service shows inactive status:

systemctl status zfs-import-cache
● zfs-import-cache.service - Import ZFS pools by cache file
   Loaded: loaded (/usr/lib/systemd/system/zfs-import-cache.service; static)
   Active: inactive (dead)

The /etc/sysconfig/zfs configuration appears standard, but these key parameters might need adjustment:

ZFS_MOUNT='yes'
ZFS_UNMOUNT='yes'
ZFS_SHARE='yes'
ZFS_UNSHARE='yes'

After thorough testing, this combination of steps resolved the issue permanently:

# 1. Create cache directory if missing
sudo mkdir -p /etc/zfs

# 2. Set cache file location
sudo zpool set cachefile=/etc/zfs/zpool.cache PoolName

# 3. Enable systemd services
sudo systemctl enable zfs-import-cache
sudo systemctl enable zfs-import.target
sudo systemctl enable zfs-mount
sudo systemctl enable zfs-share

# 4. Verify boot parameters
sudo cat /etc/default/grub | grep zfs

For environments where cache files aren't desirable, configure device scanning:

# Edit /etc/default/zfs
ZPOOL_IMPORT_PATH="/dev/disk/by-id /dev/disk/by-partuuid"

# Then regenerate initramfs
sudo dracut -f

• Check boot logs: journalctl -b | grep zfs
• Verify module loading: lsmod | grep zfs
• Test manual import: zpool import -a
• Inspect cache file: file /etc/zfs/zpool.cache