基于Debian 12 (Bookworm) 与 OpenZFS 打造可靠 Homelab 存储方案
OpenZFS 在 Linux 上已经非常稳定，早在很久以前就注意到已被纳入 Debian 的 contrib 源中。现在可以很轻松地通过 Debian 的官方源来安装 OpenZFS，享受其先进的文件系统和卷管理功能。无论是在企业级环境还是个人的 Homelab 中，OpenZFS 提供了强大的数据完整性、备份和恢复选项，使其成为理想的存储解决方案。
硬件
| 组件 | 详细信息 |
|---|---|
| 主板 | Dell Precision 3440 SFF (Intel W480) |
| CPU | Intel(R) Xeon(R) W-1270P(8C16t) |
| RAM | 32GB DDR4 |
| 网卡 | I219-LM(1G) I226-V(2.5G) |
| SSD | HFM256GDGTNG-87A0APLEXTOR PX-512M9PeG * 2 |
| HDD | WDC WD5000LPVX-22V0TT0 * 3 |
存储规划
| 存储池 | 硬盘 | 接口 | 文件系统 | RAID | 用途描述 |
|---|---|---|---|---|---|
| nvme0 | HFM256GDGTNG-87A0A | nvme | ext4 | - | 系统盘 |
| zpl-dbs | PLEXTOR PX-512M9PeG * 2 | nvme | zfs | mirror | 数据库/虚拟机 |
| zpl-shr | WDC WD5000LPVX-22V0TT0 * 3 | sata | zfs | raidz1 | 共享盘/数据备份 |
注意事项
- 配置好硬件及BIOS确保其可以断电导致关机情况下通电后可自启动
- 像Dell这样的品牌机还保留了开机检查键盘是否插好的传统需要在BIOS中禁掉
- 关闭硬件RAID功能(这里使用ZFS软RAID)
- 确保机器不会自动休眠:
sudo systemctl mask sleep.target suspend.target hibernate.target hybrid-sleep.target - 为了减少不必要软件建议安装 minimal 版本的 Debian 并且需要选择ssh服务
系统设置
-
启用sshd:
sudo systemctl enable sshd -
停用SELinux:
/etc/selinux/config->SELINUX=disabled -
网络设置: 将全部以太网卡桥接到
brlan0/etc/network/interfaces.d/brlan0.ifaceauto brlan0 iface brlan0 inet static address 10.21.0.26/24 gateway 10.21.0.1 dns-nameservers 10.21.0.1 bridge_ports eth0 eth1 bridge_stp on # bridge_ports eth0 eth1 wlan0 # pre-up wpa_supplicant -B -Dnl80211 -iwlan0 -c /etc/wpa_supplicant/wpa_supplicant.con # hostapd # pre-up hostapd -B /etc/hostapd/hostapd.conf -
安装常用软件:
apt update && apt upgradeapt install apt-utils vim sudo openssh-server sshpass openssh-client wget curl ifupdown nfs-common net-tools tcpdump tree unzip smartmontoolsapt install openssl-dev git samba nginx-full -
安装OpenZFS:
apt install zfsutils-linuxsystemctl enable --now zfs-zed.service
创建数据库存储卷
zpool create -f zpl-dbs mirror /dev/nvme0n1 /dev/nvme1n1
zpool status zpl-dbs
NAME STATE READ WRITE CKSUM
zpl-dbs ONLINE 0 0 0
mirror-0 ONLINE 0 0 0
nvme0n1 ONLINE 0 0 0
nvme1n1 ONLINE 0 0 0
创建database volume: zfs create zpl-dbs/database
可创建快照: zfs snapshot zpl-dbs/database@backup-20230612
定期备份数据: crontab -l
*/10 * * * * rsync -avz --delete /zpl-dbs/database user@<nasip>:/rsync-dbs/
共享文件系统配置
创建 zpl-shr:
zpool create -f zpl-shr raidz /dev/sda /dev/sdb /dev/sdc
zpool status zpl-dbs
NAME STATE READ WRITE CKSUM
zpl-vms DEGRADED 0 0 0
raidz1-0 DEGRADED 0 0 0
ata-WDC_WD5000LPVX-22V0TT0_WD-WX11E4434180 ONLINE 0 0 0
ata-WDC_WD5000LPVX-22V0TT0_WD-WXC1E840FKFT ONLINE 0 0 0
ata-WDC_WD5000LPVX-22V0TT0_WD-WX41A7486NSJ FAULTED 0 0 0
设置lz4压缩:
zfs set compression=lz4 zpl-shr
查看压缩率:
zfs get compressratio
NFS服务器: /etc/exports
/zpl-shr *(rw,no_root_squash,no_subtree_check)
systemctl enable --now rpcbind nfs-server
SAMBA服务: /etc/samba/smb.conf
# 免密码配置 加上 "security" 和 "map to guest"
[global]
security = user
map to guest = Bad User
[zpl-shr]
comment = Work Dir
path = /zpl-shr/
public = no
writeable = yes
browseable = yes
guest ok = yes
[user]
comment = Work Dir
path = /home/user
public = no
writeable = yes
browseable = yes
启用samba: systemctl enable smbd
添加用户: smbpasswd -a user
重启samba: systemctl restart smbd
Web服务: systemc enable --now nginx.service
/etc/nginx/sites-enabled/default
server {
listen 80 default_server;
listen [::]:80 default_server;
root /zpl-shr/www;
index index.html index.htm index.nginx-debian.html;
server_name _;
location / {
# alias /var/www/html/;
try_files $uri $uri/ =404;
}
location /files {
alias /zpl-shr/pub;
autoindex on;
autoindex_exact_size off;
autoindex_format html;
autoindex_localtime on;
}
}
其它常用服务
# cockpit
apt install cockpit cockpit-machines cockpit-podman
systemctl enable --now cockpit.socket
# podman
apt install podman
systemctl enable --now podman.service
# qemu-kvm & libvirt
apt install qemu-kvm libvirt-clients libvirt-daemon-system bridge-utils virtinst libvirt-daemon virt-manager qemu-system
systemctl enable --now libvirtd
usermod -aG kvm,libvirt,libvirt-qemu <user>
# shell in a box
apt install shellinabox
systemctl enable --now shellinabox
# vnstat
apt install vnstat
systemctl enable --now vnstat.service
ZFS常用命令
# show zfs
zpool list # list all zpool
zpool status # show zpool & disk status
zfs list # list all zfs volumes
# zpool
zpool create -f data mirror /dev/sda /dev/sdb
zpool destroy data # delete pool
zpool import -f data/tank # import zpool
# compress
zfs set compression=lz4 data
zfs get compress
zfs get compressratio
# volume
zfs create data/tank # create
zfs create -V 2gb data/tank # Create with size
zfs destroy data/tank # remove
# snapshot
zfs list -t snapshot # show all zfs snapshot
zfs snapshot data/tank@backup1 # create snapshot
zfs destroy data/tank@backup1 # delete snapshot
zfs rollback data/tank@backup1 # rollback snapshot
zfs clone data/tank@backup1 data/tank2 # clone from snapshot
zfs send data/tank@backup1|zfs receive data2/tank # copy to newpool
# zfs cache
zpool add rpool cache <your-ssd-drive>
# mount
zfs mount -a # mount all zfs
zfs unmount data/tank # unmount
zfs mount # show all mountpoint
zfs get mountpoint data/tank # get mountpoint
zfs get mounted data/tank # get mounted status
zfs set mountpoint=/my_data data/tank # set mountpoint
# Replace hard drive
ls /dev/disk/by-id/ # show disk by ID
zpool offline data disk-id1
zpool replace -f data disk-id1 disk-id2
# errors: Permanent errors have been detected in the following files:
zpool scrub data
# Setting the acltype property to posixacl indicates Posix ACLs should be used.
zfs set acltype=posixacl data
ZFS&磁盘健康检查
systemctl enable --now smartmontools.service
smartctl -i /dev/sda
smartctl --all /dev/sda
zpool scrub data # scrub zpool
zpool status -xv data # check zpool status
zpool get all data
ZVOL转换(qcow2/raw)
qemu-img convert -f qcow2 debian.qcow2 -O raw debian.raw # qcow2 to raw
zfs create -V 12gb data/vm-100-debian # create zfs volume
dd if=debian.raw of=/dev/zvol/data/vm-100-debian bs=4M # write to volume
dd if=/dev/zvol/data/vm-100-debian of=debian.raw bs=4M # write to raw