Date: Sun, 29 Jan 2017 16:34:41 -0800 From: David Christensen <dpchrist@holgerdanske.com> To: freebsd-questions@freebsd.org Subject: Re: FreeBSD 11.0-RELEASE-p7 i386 system drive imaging and migration Message-ID: <2973d1ea-202f-60fa-2930-eec05b626cfb@holgerdanske.com> In-Reply-To: <86bmupg0gi.fsf@WorkBox.homestead.org> References: <df0c81d7-fd2b-852f-4007-5fb4b24100e0@holgerdanske.com> <86bmupg0gi.fsf@WorkBox.homestead.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 01/29/17 10:55, Brandon J. Wandersee wrote:
>
> David Christensen writes:
>
>> What is the proper way to clone a FreeBSD system image from one drive to
>> another?
>
> In my personal opinion, the "proper" way is to back up your data, create
> a fresh partition table and filesystems on the new disk, and restore the
> backup. Using `dd` to clone an entire disk byte-for-byte works, but it's
> the painfully slow, tedious, and potentially dangerous way of doing
> this. The native backup utilities---dump(8) and restore(8) for UFS, `zfs
> send` and `zfs receive` for ZFS---will copy the data from an existing
> filesystem and write to a new filesystem at speeds exponentially greater
> than anything you'll get from `dd`.
Thanks for the reply.
Here's my FreeBSD system drive:
toor@freebsd:/root # gpart show
=> 63 31277169 ada0 MBR (15G)
63 1 - free - (512B)
64 31277160 1 freebsd [active] (15G)
31277224 8 - free - (4.0K)
=> 0 31277160 ada0s1 BSD (15G)
0 4194304 1 freebsd-zfs (2.0G)
4194304 4194304 2 freebsd-swap (2.0G)
8388608 22888544 4 freebsd-zfs (11G)
31277152 8 - free - (4.0K)
toor@freebsd:/root # zpool list
NAME SIZE ALLOC FREE EXPANDSZ FRAG CAP DEDUP HEALTH
ALTROOT
bootpool 1.98G 101M 1.89G - 6% 4% 1.00x ONLINE -
zroot 10.9G 4.30G 6.58G - 33% 39% 1.00x ONLINE -
toor@freebsd:/root # zfs list
NAME USED AVAIL REFER MOUNTPOINT
bootpool 101M 1.82G 99.4M /bootpool
zroot 4.30G 6.24G 96K /zroot
zroot/ROOT 2.68G 6.24G 96K none
zroot/ROOT/default 2.68G 6.24G 2.68G /
zroot/tmp 164K 6.24G 164K /tmp
zroot/usr 1.61G 6.24G 96K /usr
zroot/usr/home 399M 6.24G 399M /usr/home
zroot/usr/ports 641M 6.24G 641M /usr/ports
zroot/usr/src 609M 6.24G 609M /usr/src
zroot/var 812K 6.24G 96K /var
zroot/var/audit 96K 6.24G 96K /var/audit
zroot/var/crash 96K 6.24G 96K /var/crash
zroot/var/log 308K 6.24G 308K /var/log
zroot/var/mail 120K 6.24G 120K /var/mail
zroot/var/tmp 96K 6.24G 96K /var/tmp
As I understand it, taking an image involves:
1. Back up the MBR (dd?).
2. Back up the slice 1 partition table (?).
3. Back up bootpool file system ('zfs send').
4. Back up the swap partition encryption container header (?).
5. Back up the zroot partition encryption container header (?).
6. Back up zroot file system ('zfs send').
Restoring an image involves:
1. Restore MBR ('dd').
2. Restore slice 1 partition table (?).
3. Create bootpool ZFS pool and file system (?).
4. Restore bootpool file system ('zfs receive').
5. Create a encryption container in the second partition.
6. Restore headers.
7. Initialize swap.
8. Create an encryption container in the third partition.
9. Restore headers.
10. Create zroot ZFS pool and file system (?).
12. Restore zroot file system ('zfs receive').
(I'm a FreeBSD noob, so I'm sure there are errors in the above.)
These processes are complex enough to warrant automation. Can
Clonezilla handle FreeBSD 11.0 with MBR and encrypted ZFS root?
> However, I'm not sure that addresses the actual problem in this
> particular case. I can't say exactly what the error message you're
> getting means, but while it might stem from how you copied the system
I have a Debian 7 computer with a Pentium D 945 processor, 4 GB RAM, 16
GB USB flash system drive, a Seagate ST3000DM01 3 TB backup HDD, and an
assortment of drive docking bays.
How Debian sees the original FreeBSD system drive:
2017-01-28 13:44:37 root@p43200 ~
# parted /dev/sdb u s p free
Model: ATA SAMSUNG SSD UM41 (scsi)
Disk /dev/sdb: 31277232s
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Number Start End Size Type File system
Flags
63s 63s 1s Free Space
1 64s 31277223s 31277160s primary zfs boot
31277224s 31277231s 8s Free Space
Take an image of the original FreeBSD system drive:
2017-01-28 13:44:55 root@p43200 ~
# cd /mnt/q/image/holgerdanske.com/freebsd/
2017-01-28 13:48:05 root@p43200
/mnt/q/image/holgerdanske.com/freebsd
# time dd if=/dev/sdb count=31277224 | gzip | tee
i72600s-20170128-1346-
freebsd-11.0-release-i386-op.img.gz | sha256sum -b >
i72600s-20170128-1346-freeb
sd-11.0-release-i386-op.img.gz.sha256
31277224+0 records in
31277224+0 records out
16013938688 bytes (16 GB) copied, 1032.32 s, 15.5 MB/s
real 17m12.330s
user 12m7.961s
sys 1m42.726s
Restore image onto a cloned system drive (Intel 520 Series 60 GB SSD):
2017-01-28 20:56:58 root@p43200
/mnt/q/image/holgerdanske.com/freebsd
# time zcat
i72600s-20170128-1346-freebsd-11.0-release-i386-op.img.gz > /dev/sdb
real 3m42.036s
user 1m35.854s
sys 0m27.406s
> it might also imply a problem with the disk itself. Unrecoverable read
> errors, maybe.
1. Putting the original system drive into another computer broke Xfce
applications.
2. Everything works as before when the original system drive is put
back into original computer (I am typing this message on that system).
3. The cloned drive works and has passed Intel SSD Toolbox checks.
4. Putting the cloned drive into the original computer broke Xfce
applications in exactly the same way.
I doubt it's an SSD hardware problem.
Again, here are the error messages seen on the console when I attempt to
launch Terminal Emulator in Xfce when the original system drive is place
into another computer and when the cloned drive is placed into the
original computer:
vm_fault: pager read error: pid 1023 (python2.7)
vm_fault: pager read error: pid 1040 (xfce4-terminal)
Does anyone have any ideas why Xfce applications would generate such errors?
David
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?2973d1ea-202f-60fa-2930-eec05b626cfb>