ZFS 功能介紹

本文章轉貼自http://blog.ntut.idv.tw/2009/04/fbsd-freebsd-zfs-filesystem.html

底下為原文內容：

ZFS(Zettabyte File System) 一個打破過去思維的檔案系統，是昇陽 Sun Microsystems 這家公司所開發出來的全新型態檔案系統，因為 License 的問題所以目前只有在 Solaris、Mac、BSD 上看得到，ZFS 是 128bit 的檔案系統而它到底有多強呢？別再等待了馬上用了你就知道，只能說 ZFS 真是一個上帝賜給IT人員的好禮物。

優點：

1. 簡易管理(Ease of Management)：只需透過簡易的兩個指令 zpool 及 zfs 即可操作管理增加及縮減檔案系統容量。
2. 高延展性(Scalability)：可以線上動態的增加容量給檔案系統而不需停機。
3. 資料完整性(Data Integrity)：再也不需要 fsck 修復資料，任何在 ZFS 裡面進行的動作都會經過同位檢查(Everything is checksummed)確認後才寫入/讀出。
4. 驚人的高效能表現(Breathtaking Performance)：資料寫入磁碟機時，首先會將資料先寫入第一個未使用的區塊(first free block)，不需等待實體磁碟機轉速的延遲及磁頭移動的時間。同時具有智慧預備讀取功能(Intelligent Prefetch)可自動預測下筆讀取資料並將其預先放置在快取記憶體裡。
5. 擁有企業級的操作功能：具備 Quota(磁碟配額)、Reservation(磁碟預留)、Compression(磁碟壓縮)、Snapshot(磁碟快照)、Clone(磁碟復製)等強大的功能。
6. ZFS是免費使用檔案系統：Apple 的 Mac OS 將在下一版 10.6 Snow Leopard 即內建支援 ZFS，而 FreeBSD 7.X 環境下也已經開始支援 ZFS 也將於 8.0 之後的版本列入為主要檔案系統的一部份。

測試環境：

i386 P4-1.6
Real RAM 512M
OS FreeBSD 7.1R

1. 啟動 ZFS

# vi /etc/rc.conf #加入開機自動啟動
zfs_enable="YES"

# vi /boot/loader.conf # 針對 ZFS FreeBSD 核心的最佳化調整
vm.kmem_size="330M"
vm.kmem_size_max="330M"
vfs.zfs.arc_max="40M"
vfs.zfs.vdev.cache.size="5M"

# /etc/rc.d/zfs start # 手動啟動 ZFS
# reboot # 重新啟動讓調整的參數開機載入


2. ZFS 的 zpool 指令

# zpool create storage mirror ad2 ad3 # 建立 Raid1(1+1=1) Mirror 格式，資料會同時寫入兩棵 HD。
# df -h # 查看 storage pool 是有已經建立
Filesystem Size Used Avail Capacity Mounted on 
storage 5.8G 6.9M 5.8G 0% /storage

# zpool status # 查看本機 ZFS 的狀態，這個系統中有一個 pool 名稱為 storage 是由兩棵代號 ad2 及 ad3 的實體硬碟 mirror 而成
pool: storage
state: ONLINE
scrub: none requested
config:

NAME STATE READ WRITE CKSUM
storage ONLINE 0 0 0
mirror ONLINE 0 0 0
ad2 ONLINE 0 0 0
ad3 ONLINE 0 0 0

errors: No known data errors

# zpool list # 查看 pool 的空間使用狀態
NAME SIZE USED AVAIL CAP HEALTH ALTROOT
storage 5.94G 7.00M 5.93G 0% ONLINE -

# zpool offline storage ad2 # 讓 storage 這個 pool 的 ad2 硬碟暫時離線
Bringing device ad2 offline
# zpool status storage # 查看狀態會有 DEGRADED 警告
pool: storage
state: DEGRADED
status: One or more devices has been taken offline by the administrator.
Sufficient replicas exist for the pool to continue functioning in a
degraded state.
action: Online the device using 'zpool online' or replace the device with
'zpool replace'.
scrub: none requested
config:

NAME STATE READ WRITE CKSUM
storage DEGRADED 0 0 0
mirror DEGRADED 0 0 0
ad2 OFFLINE 0 0 0
ad3 ONLINE 0 0 0

errors: No known data errors

# zpool online storage ad2 # 讓 storage 這個 pool 的 ad2 硬碟重新上線
Bringing device ad2 online
# zpool status storage # 查看狀態 DEGRADED 警告已解除
pool: storage
state: ONLINE
scrub: resilver completed with 0 errors on Thu Apr 9 13:49:50 2009
config:

NAME STATE READ WRITE CKSUM
storage ONLINE 0 0 0
mirror ONLINE 0 0 0
ad2 ONLINE 0 0 0
ad3 ONLINE 0 0 0

errors: No known data errors

# zpool export -f storage # 卸除(umount) storage 這個 pool 參數 -f (Force) 為強制卸除
# df -h # 查看是否 umount，已無 storage 這個 pool
Filesystem Size Used Avail Capacity Mounted on
/dev/ad0s1a 11G 1.4G 8.5G 14% /
devfs 1.0K 1.0K 0B 100% /dev

# zpool import storage # 掛載(mount) storage 這個 pool
# df -h # 查看是否 mount，已重新掛載 storage pool
Filesystem Size Used Avail Capacity Mounted on
/dev/ad0s1a 11G 1.4G 8.5G 14% /
devfs 1.0K 1.0K 0B 100% /dev
storage 5.8G 6.9M 5.8G 0% /storage

# zpool scrub storage # 手動驗證 storage 這個 pool 中所有資料的完整性
# zpool replace storage ad4 ad5 # 在 storage pool 中用新的 ad4 ad5 取代原本的 ad2 ad3 兩棵 HD


3. ZFS 的 zfs 指令

# zfs create storage/compressed # 在 storage 這個 pool 中建立一個檔案目錄叫 compressed
# zfs list # 查看 ZFS 檔案狀態
NAME USED AVAIL REFER MOUNTPOINT
storage 7.02M 5.84G 6.91M /storage
storage/compressed 18K 5.84G 18K /storage/compressed

# zfs set compression=gzip storage/compressed # 將 compressed 目錄設定成 gzip 壓縮格式
# zfs set compression=off storage/compressed # 解除 compressed 目錄設的壓縮格式

# zfs mount # 查看所有 ZFS 掛載的磁區
storage /storage
storage/compressed /storage/compressed

# zfs get all storage/compressed # 查看 storage/compressed 檔案目錄屬性
NAME PROPERTY VALUE SOURCE
storage/compressed type filesystem -
storage/compressed creation Thu Apr 9 14:14 2009 -
storage/compressed used 18K -
storage/compressed available 5.84G -
storage/compressed referenced 18K -
storage/compressed compressratio 1.00x -
storage/compressed mounted yes -
storage/compressed quota none default
storage/compressed reservation none default
storage/compressed recordsize 128K default
storage/compressed mountpoint /storage/compressed default
storage/compressed sharenfs off default
storage/compressed checksum on default
storage/compressed compression gzip local
storage/compressed atime on default
storage/compressed devices on default
storage/compressed exec on default
storage/compressed setuid on default
storage/compressed readonly off default
storage/compressed jailed off default
storage/compressed snapdir hidden default
storage/compressed aclmode groupmask default
storage/compressed aclinherit secure default
storage/compressed canmount on default
storage/compressed shareiscsi off default
storage/compressed xattr off temporary
storage/compressed copies 1 default

# zfs set quota=3G storage/compressed # 可設定檔案空間大小為 3G 但 storage 這個 pool 總大小為 5.8G 如果 3G 滿了而 5.8 還未用完會 share 給 /storage/compressed 繼續使用
# df -h # 查看 storage/compressed 的 Avail 變成 3G
Filesystem Size Used Avail Capacity Mounted on
storage 5.8G 7.0M 5.8G 0% /storage
storage/compressed 3.0G 128K 3.0G 0% /storage/compressed

# zfs set reservation=3g storage/compressed # 設定保留了 3G 給 /storage/compressed
# df -h # 查看 storage 的 Size 變成 2.8G，這是因為保留了 3G 給 /storage/compressed，代表 /storage/compressed 真正擁有 3G 別的 filesystem 是不能跟它 share 的
Filesystem Size Used Avail Capacity Mounted on
storage 2.8G 7.0M 2.8G 0% /storage
storage/compressed 3.0G 128K 3.0G 0% /storage/compressed

# zfs create storage/data # 於 storage pool 中再建立一個 data 檔案目錄
# zfs set mountpoint=/data storage/data # 設定真實掛載點原本 Mounted on 會由 /storage/data 變成到 /data
# df
Filesystem 1K-blocks Used Avail Capacity Mounted on
/dev/ad0s1a 11294270 1468766 8921964 14% /
devfs 1 1 0 100% /dev
storage 2982656 7040 2975616 0% /storage
storage/compressed 3145728 128 3145600 0% /storage/compressed
storage/compressed@2009-04-09 3145728 128 3145600 0% /storage/compressed/.zfs/snapshot/2009-04-09
storage/data 2975616 0 2975616 0% /data

# zfs set sharenfs=rw storage/data # 設定分享權限可讀及寫，通常用在 NFS 分享上

# touch /storage/compressed/1234 # 建一個檔案到 compressed 目錄下
# md5 /storage/compressed # 用 md5 去 hash snapshot 之前的目錄，會得到一個 hash 值
MD5 (/storage/compressed) = 7bffed2808dfba7915f89f8f42b09f83
# zfs snapshot storage/compressed@2009-04-09 # snapshot compressed 目錄，取個隨便名稱就用今天日期
# md5 /storage/compressed/.zfs/snapshot/2009-04-09 # 再次用 hash 去驗證 snapshot 前後所得到的檔案是否一樣
MD5 (/storage/compressed/.zfs/snapshot/2009-04-09) = 7bffed2808dfba7915f89f8f42b09f83

# zfs destroy storage/compressed # 可刪除 storage/compressed 檔案目錄

LVM 64bit

All filesystems have limits, and the Ext family is no exception. Using your standard 4KiB block size, the 32-bit Ext3 has a well-known limit of 16TB for a volume. With the advance of storage mediums and greater availability of increased capacity, Ext4 was developed to overcome this limitation. Ext4 will now support volumes up to 1 exbibyte (EiB).

This is a fantastic, improvement, but what happens when you actually go to format one of these volumes? If your system is running with some of the older fs tools (like Debian Squeeze), then you will probably encounter this scenario when trying to format a large volume (let's say we have a 19TB partition at /dev/vg0/lv_data for this example).

# mkfs -t ext4 /dev/vg0/lv_data mke2fs 1.42.7 (21-Jan-2013) mkfs.ext4: Size of device (0x131a47800 blocks) /dev/vg0/lv_data too big to be expressed in 32 bits using a block-size of 4096.

Although the Ext4 filesystem has been updated to support 64-bit volumes, the tools may not be. The solution here is to grab the latest version of the e2fsprogs utilities and build them yourself.

First, grab the utilities from sourceforge:

# cd /opt/
# wget -Oe2fsprogs-1.42.7.tar.gz 
http://downloads.sourceforge.net/project/e2fsprogs/e2fsprogs/v1.42.7/e2fsprogs-1.42.7.tar.gz?r=http%3A%2F%2Fe2fsprogs.sourceforge.net%2F&ts=1361548232&use_mirror=superb-dca3

Then, extract the files and setup for your build

# tar -xzvf e2fsprogs-1.42.7.tar.gz
# cd e2fsprogs-1.42.7
# mkdir build
# cd build

If you don't have dev tools (like a compiler and make), install them now # apt-get install build-essential

Now you are ready to build the tools:

# ../configure
# make
# make install

Before actually creating your filesystem, however, you are going to wan to edit your /etc/mke2fs.conf file and enable the 64-bit feature flag automatically for a big disk. It should look something like this:

[fs_types]
    ext4 = {
        features = has_journal,extent,huge_file,flex_bg,uninit_bg,dir_nlink,extra_isize
        auto_64-bit_support = 1 # <—- add this line
        inode_size = 256
    }

Finally, you are ready to create your volume! If you are having trouble with the standard tools, you can run this to manually specify your options: # mke2fs -O 64bit,has_journal,extents,huge_file,flex_bg,uninit_bg,dir_nlink,extra_isize -i 4194304 /dev/vg0/lv_data

Now, you should be able to mount your volume and use it!

# mount /dev/vg0/lv_data /mnt
# cd /mnt