2. Business Continuity Plans and Backups
•
People make IT support a complex issue!
•
Disaster Recovery must be based on Business
Continuity Plans and meet the requirements as
set in the following question:
What is the cost of downtime per hour?
March 1, 2014
Webinar
2
3. Loss of Data - Most Feared Threat
Human error
35
Systems failure
31
% o f re s p o nd e nts
Supply chain disruption
29
Virus, worm or other malicious attack on IT systems
28
Employee malfeasance (e.g. theft or fraud)
25
Natural disasters, such as fires or floods
22
Unplanned downtime of online systems
22
Terrorism
16
Power outage
13
Pandemic
13
Application failure
Industrial Action
March 1, 2014
12
8
Webinar
3
4. Bootable System Images in Unix and
Linux
Many tools available. For the sake of brevity,
the following will be discussed:
AIX
mksysb, Network Installation Manager (NIM)
HP
make_tape_recovery/make_net_recovery,
Dynamic Root Disk (DRD)*
Linux
Mondo Rescue, Clonezilla
Solaris ufsdump, fssnap+ufsdump, flash/JumpStart
Tru64 btcreate
March 1, 2014
Webinar
4
5. Tape Drives
Limitations inherent with tape media:
•
A tape drive must be available on each system to be archived.
•
Must remove old tapes and insert new ones for new backups.
•
If an archive exceeds the capacity of a tape, you must swap tapes for
both creation and extraction.
•
Must check log files and run dummy restores to ensure data
consistency.
•
Tape drives are more error-prone than a local network or CD-ROM
and DVD.
•
Cost of tapes in large environment is significant.
•
Cost of managing tape loading and storage is significant.
•
Generally slower that disk or LAN.
March 1, 2014
Webinar
5
6. AIX – mksysb(1)
Creates a backup of the operating system (root
volume group).
The file system image is in backup-file format. The
tape format includes a boot image, a bosinstall
image, and an empty table of contents followed by
the system backup (root volume group) image.
The root volume group image is in backup-file
format, starting with data files and then any
optional map files.
March 1, 2014
Webinar
6
7. AIX – mksysb(1)
Pros highlights:
• For managing a single or limited number of
servers.
• Can be used for system cloning.
• Use when servers are not networked.
• Allows backup to tape drive (CD-ROM and DVD
typically done through mkcd(1)).
March 1, 2014
Webinar
7
8. AIX – mksysb(1)
Cons highlights:
• Identical tape drive is needed for an off-site recovery.
• Cannot back up files that are mounted from a remote server.
• If /usr is remote-mounted, you cannot reinstall system from backup
image.
• Image does not include data on raw devices or in user-defined paging
spaces.
• It may not restore all device configurations for special features, such
as /dev/netbios and some device drives not shipped with the product.
• LC_ALL environment variable should be unset (if non-C value).
• Does not have built-in error checking to minimize problems when
backing up an active file system.
• Format specific to AIX (backup-file).*
March 1, 2014
Webinar
8
10. AIX – mksysb(1) Recovery
•
Boot off the tape drive.
•
Select option 3 “Maintenance mode for system
recovery”.
•
Access your devices.
•
Restore files.
•
Boot.
March 1, 2014
Webinar
10
11. AIX – NIM(1)
•
Requires a NIM master and the number of client
instances.
•
NIM depends on certain protocols (NFS, bootp or
DHCP, and TFTP). Older versions of AIX also
required RSH and other RCMD commands, but in
AIX 5.3 and above you can use basic nimsh or
openssl.
•
NIM master must be at the highest level of AIX
that it is required to support.
•
Now supports Linux installations too.
March 1, 2014
Webinar
11
12. HP-UX – make_tape_recovery(1)
Pros highlights:
• For managing a single or limited number of servers.
• Can be used for system cloning.
• Use when servers are not networked.
• Suitable tape drive must exist.
• Allows backup to tape drive, CD-ROM, or DVD.
• Interactive and non-interactive.
• Supports tar (default), cpio or pax formats.
• Supports LVM and VxVM.
• Multi-tape CLUI only.
• Configurable.
• Versions 7.x and above have ability to block particular
paths and protocols during inventory (see instl_adm(4)).
March 1, 2014
Webinar
12
13. HP-UX – make_tape_recovery(1)
Cons highlights:
•
Identical tape drive is needed for an off-site recovery.
•
Cannot back up files that are mounted from a remote
server.
•
Does not have built-in error checking to minimize
problems when backing up an active file system.
•
LVM disk mirrors not restored.
•
LVM physical extents allocated to a logical volume may
be in a different location on a disk than before (consider
extending contiguous volumes).
•
Cannot use remote tape drive.*
March 1, 2014
Webinar
13
15. HP-UX – make_net_recovery(1)
Pros highlights:
•
For managing a large number of servers.
•
Can be used for system cloning.
•
Use when servers are networked.
•
No tape, CD-ROM, or DVD drives needed.
•
Interactive and non-interactive.
•
Supports tar (default), cpio or pax formats.
•
Highly configurable.
March 1, 2014
Webinar
15
16. HP-UX – make_net_recovery(1)
Cons highlights:
•
Requires large space if many clients are backed
up.
•
Ignite-UX bundles must be at same version on
server and clients.
•
NFS used to save data to Ignite server (firewall
issues, especially older versions of NFS)*.
•
Requires Ignite server for recoveries.
March 1, 2014
Webinar
16
17. HP-UX – make_net_recovery Protocols
and Ports
67
bootpd UDP2 Bootstrap Protocol Server This service should function only if the server is a BOOTP/DHCP server
68
bootpd UDP Bootstrap Protocol Client This service should function only if the server is a BOOTP server
69
tftpd
UDP Trivial File Transfer Protocol Found on systems that have Ignite-UX installed. This service should
function only if the host is being used as a TFTP server
111
portmap/sunrpc/ rpcbind
Procedure Call (RPC)
March 1, 2014
TCP/UDP
Webinar
SUN Remote
17
18. HP-UX – make_net_recovery Protocols
and Ports (continued)
135 rpcd/dced
TCP Distributed Computing
Environment (DCE)-based RPC
514
shell TCP
Remote Command, No Password Used
1067 instl_boots
Server
UDP Installation Bootstrap Protocol
1068 instl_bootc
Client
UDP Installation Bootstrap Protocol
2049 nfsd
March 1, 2014
TCP/UDP
NFS Remote File System
Webinar
18
19. HP-UX – make_net_recovery Protocols
and Ports (continued)
2121 swagentd
TCP/UDP
HP Software Distributor
Daemon - Used for communication between systems for
software installation, listing, or other sw commands
4000 - 4009 secure swagent ports
TCP/UDP
The swagent firewall configurable ports
49152 - 65535
Dynamic or Private Ports TCP/UDP
Dynamic and Private Ports are used by many applications
for
dynamic port assignments. UDP ports in this range are
often
RPC ports
March 1, 2014
Webinar
19
21. HP-UX – Dynamic Root Disk
•
DRD (current release A.3.1.0 - February 2008) runs on
both Integrity and PA platforms running the following
operating systems:
HP-UX 11i v2 (11.23) September 2004 or more recent
HP-UX 11i v3 (11.31)
•
Root group being cloned can be managed by any release
of LVM on an O/S release supported by DRD. In addition,
the root group can be managed by VxVM 4.1 (HP-UX 11i
v2 or 11i v3) or VxVM 5.0 (HP-UX 11i v2 only).
March 1, 2014
Webinar
21
22. HP-UX DRD Benefit: Minimizing
Planned Downtime
Without DRD: Software management may require extended downtime
With DRD: Install/remove software on the clone while applications continue running
Install patches
on the clone;
applications
remain running
lvol1
lvol2
lvol3
lvol1
lvol2
lvol3
boot disk boot mirror
vg00 (active)
Activate the
clone to make
changes take
effect
lvol1
lvol2
lvol3
lvol1
lvol2
lvol3
boot disk boot mirror
vg00 (inactive)
March 1, 2014
Webinar
lvol1
lvol2
lvol3
lvol1
lvol2
lvol3
clone disk clone mirror
cloned vg00 (inactive/patched)
lvol1
lvol2
lvol3
lvol1
lvol2
lvol3
clone disk clone mirror
cloned vg00 (active/patched)
22
23. HP-UX – Dynamic Root Disk
Pros highlights:
•
Fully supported by HP.
•
Full clone.
•
Complements other parts of total HP solution by reducing system
downtime required to install and update patches and other software.
•
Copy operation is currently done by fbackup and frecover.
•
Kctune(1) command can be used to modify kernel parameters in the
clone.
•
The ioconfig file and the entire /dev directory are copied by the DRD
clone operation, so instance numbers will not change when the clone
is booted.*
•
Supports nPars, vPars, and Integrity VMs.
March 1, 2014
Webinar
23
24. HP-UX – Dynamic Root Disk
Pros highlights:
•
No tape drive is needed.
•
No impact on network performance.
•
No security issues of transferring data across the network.
•
All DRD processes, including drd clone and drd runcmd,
can be safely interrupted issuing Control-C (SIGINT) from
the controlling terminal or by issuing kill -HUP<pid>
(SIGHUP). This action causes DRD to abort processing
and perform any necessary clean up. Do not interrupt
DRD using the kill -9 <pid> command (SIGKILL), which
fails to abort safely and does not perform cleanup.
March 1, 2014
Webinar
24
25. HP-UX – Dynamic Root Disk
Cons highlights:
•
VxVM 5.0 not supported in HP-UX 11.31 yet.
•
Target disk must be a single disk.
•
Not easy to list all differences between active VG and the
clone.**
•
Cloning should be done when the server’s activity is
quiescent.
•
Cloned disk can be used to boot another system - it is
possible to do this, however, factors such as machine
personality (e.g., hostname, IP address and so on) make
this very difficult. HP does not recommend using the
cloned disk to boot another system.
March 1, 2014
Webinar
25
26. HP-UX – Dynamic Root Disk
Cons:
•
Only the contents of vg00 are copied. A system that has /opt (or any
file system that is patched) not in vg00 is not suitable for use with
DRD.
•
Does not provide a mechanism for resizing file systems during a drd
clone operation. However, after the clone is created, you can
manually change file system sizes on the inactive system without
needing an immediate reboot. The whitepaper, Using the Dynamic
Root Disk Toolset describes resizing file systems other than /stand.
The whitepaper Using the DRD toolset to extend the /stand file
system in an LVM environment describes resizing the boot (/stand)
file system on an inactive system image.
•
Current release of DRD does not copy the Itanium service partition
(s3).
March 1, 2014
Webinar
26
27. HP-UX – Dynamic Root Disk Examples
HP-UX 11.21:
# drd clone -t /dev/dsk/c2t1d0 -x overwrite=true [-x
mirror_disk=/dev/dsk/c3t0d1]
HP-UX 11.31, use agile views:
# drd clone -t /dev/disk/disk32 -x overwrite=true [-x
mirror_disk=/dev/disk/disk41]
Note that all partitions on Itanium disk are created
and s1 and s2 are copied.
March 1, 2014
Webinar
27
29. Linux – Mondo Rescue
Mondo Rescue is a GPL disaster recovery
solution.
It supports Linux (i386, x86_64, ia64) and
FreeBSD
(i386).
Packaged for multiple distributions (RedHat,
RHEL,
SuSE, SLES, Mandiva, Debian, Gentoo).
March 1, 2014
Webinar
29
30. Linux – Mondo Rescue
Pros highlights:
•
GNU General Public License (GPL).
•
Supports LVM 1 and 2, RAID, ext2, ext3, JFS, XFS,
ReiserFS, VFAT and UFS.
•
Supports tapes, disks, network and CD/DVD as backup
media, multiple file systems, USB key/disks, LVM,
software and hardware RAID (no more floppy support).
•
Can use used in interactive and non-interactive mode.
•
Can backup data to NFS.
•
Can move/resize/re-allocate partitions.
•
Supports GRUB and LILO boot managers.
March 1, 2014
Webinar
30
31. Linux – Mondo Rescue
Cons highlights:
• Certain packages can create problems.
• Watch our for number of free inodes (backups can fail).
• Long backups due to huge sparse /var/log/lastlog file.
• Cannot handle system and hidden attributes when archiving DOS/Windows files.
• Number of bugs: http://trac.mondorescue.org/
• Does not support Red Hat GFS yet.
• Does not support multipathing devices (/dev/mapper/mapthXpY).
• Should have option to select NIC when archiving across network (currently it takes the
first interface available).
• Pre- and Post-install script support for restores.
• Does not support bootable USB that backs up to DVDs.
• Does not support sshfs (CloneZilla has it) and webdav.
• Can hang if no floppy present.
• Issues when mounting /proc in a chroot environment (for example, mount –bind /proc
/var/named/chroot/proc).
• Does not work with SELinux.
March 1, 2014
Webinar
31
33. Linux – Mondo Rescue Network
Recovery
•
Uses network by default.
•
Boot from the Mindi mondorescue.iso and hit ENTER a
few times to restore. Mindi (Mindi-Linux) makes a minidistribution from your kernel, modules, modules, tools and
libraries. It can also generate an El Torito 2.88/5.76MB
boot disk image. Mondo uses Mindi to create a minidistro, then boots from it and runs on it.
•
The ISO images can also be used for a PXE restore. For
this to work, refer to the file README.pxe provided with
Mindi package.
March 1, 2014
Webinar
33
34. Linux – Clonezilla
Clonezilla is a GPL disaster recovery solution.
It supports Linux and Microsoft Windows.
Clonezilla Live: Allows you to use CD/DVD or USB flash drive to boot
and run clonezilla (unicast only).
Clonezilla server edition: A DRBL* server must first be set up in order to
use Clonezilla (Both unicast and multicast are supported).
Based on Partimage, ntfsclone and dd to clone partition. However,
clonezilla, containing some other programs, can save and restore not
only partitions, but also a whole disk.
March 1, 2014
Webinar
34
35. Linux – Clonezilla
Pros highlights:
•
•
•
File system supported: ext2 2, ext 3, ReiserFS, XFS, JFS,
FAT, and NTFS. For these file systems, only used blocks
in partition are saved and restored. For unsupported file
system, sector-to-sector copy is done by dd in Clonezilla.
LVM 2 is supported.
Multicast is supported in Clonezilla server edition, which is
suitable for massive cloning. You can also remotely use it
to save or restore machines if PXE and Wake-on-LAN are
supported in your clients.
March 1, 2014
Webinar
35
36. Linux – Clonezilla
Cons highlights:
•
LVM 2 is supported but LVM 1 is not.
•
Multicast is supported in Clonezilla server edition,
which is suitable for massive cloning. You can
also remotely use it to save or restore machines if
PXE and Wake-on-LAN are supported in your
clients.
•
Due to the limitations of program mkisofs , ocsiso can not process an image file larger than 4.5
GB. For this reason, if your image is larger than
4.5 GB, ocs-iso will refuse to process it.
March 1, 2014
Webinar
36
37. Linux – Clonezilla Examples
# /opt/drbl/sbin/ocs-live
# /opt/drbl/sbin/ocs-iso myimage*
# /opt/drbl/sbin/ocs-iso -g en -t -k NONE -e "-g auto -b -c restoredisk
myimg2 hda" myimg2**
# /opt/drbl/sbin/ocs-iso image3 image4
# /opt/drbl/sbin/ocs-iso -g en -k NONE -s -m ./cust-ocs***
# /opt/drbl/sbin/ocs-live-dev -g en -k NONE -s -c -m ./cust-ocs***
March 1, 2014
Webinar
37
38. Solaris – ufsdump(1)
Pros highlights:
•
Easier to restore individual files.
•
Allows you to choose directories to back up.
•
Allows you to back up entire system.
•
Allows backup to tape drive (both local and
remote!), CD-ROM, file, or diskette.
•
Supports UFS and VxFS.
•
Portable to other versions of dump/restore.
•
Interactive and non-interactive.
March 1, 2014
Webinar
38
39. Solaris – ufsdump(1)
Cons highlights:
• Backs up single file system only – you must enter multiple
ufsdump commands to back up data.
• Possible errors when backing up open files.*
• Cannot automatically calculate number of tapes need.**
• Cannot back up files that are mounted from remote
server.
• Does not have built-in error checking to minimize
problems when backing up an active file system.***
• Not efficient in tuning transfer rates to tape drive.
• Ufsrestore requires some prior knowledge of disk
partitioning.
• Does not support ZFS.
March 1, 2014
Webinar
39
40. Solaris – ufsdump(1) Comparison with
HP-UX Ignite
•
make_tape_recovery creates a bootable tape.
There is no need to boot of the installation CDROM or DVD.
•
make_tape_recovery does not require to partition
the boot disk manually in recovery process.
•
make_tape_recovery is fully automated.
•
Solaris ufsdump resembles fbackup in HP-UX.
March 1, 2014
Webinar
40
42. Solaris – ufsrestore(1) Example
1.Boot from the media at OBP prompt:
ok boot -s cdrom
2. Format the new boot disk.
3. Newfs each of the partitions on the boot disk
that are to be restored:
# newfs /dev/rdsk/c0t0d0s0
# newfs /dev/rdsk/c0t0d0s3
# newfs /dev/rdsk/c0t0d0s4
# newfs /dev/rdsk/c0t0d0s5
(/)
(/usr)
(/var)
(/tmp)
4. Each slice should be fsck to make sure newfs
worked.
March 1, 2014
Webinar
42
43. Solaris – ufsrestore(1) Example
(continued)
5. Mount all slice to be restored (do not mount or restore
swap):
# mount /dev/dsk/c0t0d0s0 /a
# cd /a
# ufsrestore rvf /dev/rmt/0n
# rm restoresymtable
# mount /dev/dsk/c0t0d0s3 /a/usr
# mount /dev/dsk/c0t0d0s4 /a/var
# mount /dev/dsk/c0t0d0s5 /a/tmp …
# cd /a/usr
# ufsrestore rvf /dev/rmt/0n …
March 1, 2014
Webinar
43
44. Solaris – ufsrestore(1) Example
(continued)
6. Unmount the file systems:
# cd /
# umount /a/usr
# umount /a/var …
7. For Solaris 2.5 and greater, run the installboot(1)
program to re-install the boot block:
# cd /usr/platform/`uname -i`/lib/fs/ufs
# installboot bootblk /dev/rdsk/c0t0d0s0
8. Check the file systems:
# fsck /dev/rdsk/c0t0d0s0
# fsck /dev/rdsk/c0t0d0s3 ...
9. Reboot the server.
March 1, 2014
Webinar
44
45. Solaris – flash(1)
Pros highlights:
•
Creates archives in cpio (default) or pax formats.
Use “-L pax” for flarcreate(1) if individual files are
larger than 4 GB.
•
Can create differential archives (flag “-A
unchanged_master_image_dir”).
•
Customisable.
•
Flash archives can be copied to NFS, HTTP or
HTTPS server, FTP server, tape, CD-ROM, DVD,
diskette, and local drive of clone system.
•
Non-interactive.
March 1, 2014
Webinar
45
46. Solaris – flash(1)
Cons highlights:
•
Flash creation software removes all RAID-1 volume
information from the archive to keep the integrity of the
clone system.
•
VxVM stores configuration information in areas not
available to Solaris Flash. If VxVM file systems have been
configured, do not create Flash archives.
•
Active sockets (like /var/tmp/orbit-* directories) can cause
flash failures).
•
The master system and the clone systems must have the
same kernel architectures.
•
Soft partitions not handled properly.
March 1, 2014
Webinar
46
48. Solaris – JumpStart Protocols and Ports
TCP and UDP 37 (time)
UDP 67 (bootp/DHCP)
UDP 69 (TFTP)
TCP and UDP 2049 (NFSv4)
TCP and UDP 4045 (lockd, may not be needed)
UDP 111 (Sun RPC)
March 1, 2014
Webinar
48
50. Solaris – flash(1) Restore Example
•
If you want to install the system using a flash archive,
select Initial option. Follow the prompts and answer
questions.
•
Options offered:
F2_Upgrade F3_Go Back F4_Initial F5_Exit F6_Help
Select F4_Initial.
•
Select Solaris Interactive Installation (Menu 2). Follow the
prompts and answer questions.
March 1, 2014
Webinar
50
51. Solaris – flash(1) Restore Example
(continue)
•
Options offered:
F2_Standard F3_Go Back F4_Flash F5_Exit F6_Help
Select F4_Flash and follow the prompts:
Flash Archive Installation Method
Available Retrieval Methods
[ ] HTTP -> default
[ ] NFS
[ ] Local File
[X] Local Tape -> selected
[ ] Local Device
March 1, 2014
Webinar
51
52. Solaris – flash(1) Restore Example
(continue)
F2_Continue F5_Cancel F6_Help
Preserve Data?
F2_Continue F3_Go Back F4_Preserve F5_Exit F6_Help
File System and Disk Layout
F2_Continue F3_Go Back F4_Customize F5_Exit F6_Help
Mount Remote File Systems?
F2_Continue F3_Go Back F4_Remote Mounts F5_Exit F6_Help
March 1, 2014
Webinar
52
54. Solaris – Bootable JumpStart Installation
CD-ROM
http://www.sun.com/blueprints/0301/BuildBoot.pdf
http://mah.everybody.org/docs/bootable-cdrom-for-solaris/
March 1, 2014
Webinar
54
55. Solaris – fssnap(1)
Create a snapshot of a file system:
The block special device created for the snapshot
is
/dev/fssnap/0
# fssnap -F ufs -o backing-store=/var/tmp/snap /home
/dev/fssnap/0
NOTE: Do not use tmpfs (/tmp) for backing store!
March 1, 2014
Webinar
55
56. Solaris – fssnap(1) (continued)
Backing up a file system snapshot without
unmounting the file system:
Since ufsdump requires the path to a raw device,
the raw option is used. The /home file system
snapshot is then removed
# ufsdump 0uf /dev/rmt/0 `fssnap -F ufs -o
raw,bs=/dev/rdsk/c2t1d0s6 /home
# fssnap -F ufs -d /home
March 1, 2014
Webinar
56
57. Solaris – fssnap(1) (continued)
Backing up a file system:
When backing up a file system, do not let the
backing-store file exceed 400 Mbytes. The second
command removes the /home file system snapshot
# ufsdump 0uf /dev/rmt/0 `fssnap -F ufs -o
maxsize=400m,backing-store=/var/tmp/snap,raw /home`
# fssnap -F ufs -d /home
March 1, 2014
Webinar
57
58. Solaris – fssnap(1) (continued)
Incremental dump of a file system:
# ufsdump IfNu /dev/rmt/0 /dev/rdsk/c0t3d0s2
`fssnap -F ufs -o raw,bs=/var/tmp/scratch,unlink
/dev/rdsk/c0t3d0s2`
March 1, 2014
Webinar
58
59. Solaris – fssnap(1) (continued)
Listing available snapshots:
# fs s na p -i
0 /va r
1 /ho m e
2 /us r/lo c a l
March 1, 2014
Webinar
59
60. Solaris – fssnap(1) (continued)
Display snapshot details:
# fs s na p -i -o ba c king -s to re -le n, ba c king -s to re , c re a te tim e
/ho m e
1 9 6 6 0 8 /va r/tm p /s na p 2 Thu Fe b 1 3 1 6 : 3 5 : 2 8 2 0 0 8
March 1, 2014
Webinar
60
61. Solaris – fssnap(1) (continued)
Mount file system snap:
Create a file system snapshot. Then, mount it on
/tmp/myfs for temporary read-only access
# fssnap -F ufs -o backing-store=/var/tmp/scratch /home
/dev/fssnap/1
# mkdir /somefilesystem/myfs
# mount -F ufs -o ro /dev/fssnap/1 /somefilesystem/myfs
March 1, 2014
Webinar
61
62. Solaris Containers – Flash Archives
•
All zones must be stopped when the flash archive
is made from the global zone.
•
If the source and target systems use different
hardware configurations, device assignments
must be changed after the flash archive is
installed.
•
Soft partitions in SVM cannot be flash archived
yet.
March 1, 2014
Webinar
62
63. Tru64 – btcreate(1)
Pros highlights:
•
Can create bootable Standalone System (SA)
kernel on tape for UFS and ADVFS.
•
Interactive and non-interactive backups.
Cons highlights:
•
Specific to Tru64.
•
Identical tape drive is needed for an off-site
recovery.
March 1, 2014
Webinar
63
64. Tru64 – btcreate(1)
The tape consists of a tape boot block, a specialized
kernel, a stripped-down operating system, and a backup
of your system partitions. Once it is created, complete
restores are as simple as booting the prepared tape from
the console:
>>> init
>>> show dev
>>> boot -fl "nc" MKA400
• Because the facility utilizes dump or vdump, the normal
caveats apply. Ideally, you should create the tape while
the system is in single-user mode:
•
# /usr/sys/bin/btcreate -f -k MYKERNEL -m mfs -t nrmt1h
-s /nfsdrive/btcreate.mykernel
March 1, 2014
Webinar
64
65. Tru64 – Bootable CD for NHD7 Tru64
5.1B on Alpha DS 15 server*
•
•
•
Copy the files from the CD in a directory ("BUILD“)
Add the "ds15kernel" file
Then run the following Tru64 command's:
# cd BUILD
# mkisofs -D -R -d -o -b -quiet -p "your name"
-P "your org." -V "V5.1Br2650_O1" -o ../TRU64DS15.iso .
# disklabel -w -t cdfs -f ../TRU64DS15.iso
# echo "0c" || dd bs=1024k conv=sync >> ../TRU64DS15.iso
Copy the ISO to a PC with a CD burner and burn the image to CD-R
• Put the cd into your DS15, and boot it using:
P>>> b -file ds15kernel dqa0
•
March 1, 2014
Webinar
65
66. Bare Metal Recovery – Selected
Commercial Products
•
HP Data Protector Express offers extended platform
support with six different Bare Metal Disaster Recovery
methods (integrated into the product). Supports Linux,
Windows, and NetWare.*
•
Symantec (Veritas) Bare Metal Restore. Part of Enterprise
NBU 6.5. Includes support for AIX, HP-UX, Solaris, Linux,
and Windows (why would anyone use the latter?).
•
IBM Tivoli Storage Manager for System Backup and
Recovery (for AIX 5.x only). Supports full system
(installation image), volume group, file system (JFS,
JFS2, NFS, CDFS), file or directory, and raw logical
volume.
March 1, 2014
Webinar
66
67. Bare Metal Recovery –Commercial
Products
•
EMC Home Base.
March 1, 2014
Webinar
67
68. Bare Metal Recovery –Commercial
Products
•
UniTrends Data Protection Unit (appliance).
March 1, 2014
Webinar
68
69. Where to Find More Information
•
AIX mksysb:
http://publib.boulder.ibm.com/infocenter/clresctr/vxrx/topic/c
om.ibm.cluster.csm16010.install.doc/am7il_mksysb.html
•
AIX script to automate mksysb via NIM:
http://www.unix.com/shell-programming-scripting/47006script-automate-mksysb-via-nim-aix-5-3-a.html
•
AIX FAQ:
http://www.faqs.org/faqs/aix-faq/
•
Symantec (Veritas) Bare Metal Restore:
http://www.symantec.com/products/
March 1, 2014
Webinar
69
70. Where to Find More Information
•
HP-IX Ignite:
http://docs.hp.com/en/IUX/
•
Tru64 btcreate:
http://h30097.www3.hp.com/docs/base_doc/DOCUMENTA
TION/V51B_HTML/MAN/MAN8/0043____.HTM
•
Tru64 bare metal recovery:
http://www.backupcentral.com/components/com_mambowi
ki/index.php/Tru64_Bare_Metal_Recovery
•
Backup Central:
http://www.backupcentral.com/
March 1, 2014
Webinar
70
71. Where to Find More Information
•
EMC HomeBase:
http://www.emc.com/solutions/samples/backup-recoveryarchiving/bare-metal-recovery.htm
•
HP Data Protector Express:
http://www.hp.com/go/dataprotectorexpress
•
Mondo Rescue:
http://www.mondorescue/org/
•
Mondo Rescue hardware migration:
http://www.mondorescue.com/docs/hwmigration-2.3.pdf
•
Solaris Flash:
http://www.sun.com/bigadmin/content/submitted/flash_archive.jsp
March 1, 2014
Webinar
71
72. Where to Find More Information
•
Diskless Remote Boot in Linux:
http://drbl.sourceforge.net/
•
Clonezilla:
http://clonezilla.sourceforge.net/
•
HP-UX Dynamic Root Disk:
http://docs.hp.com/en/DRD/faq.html
March 1, 2014
Webinar
72
Notes de l'éditeur
My humble attempt to summarise best-known utilities at the present time.
Even after 23 years of Unix experience I cannot claim I know everything!
It is possible to backup only a portion of the files (and other objects in the case of a backup) on a server.
There are three types of backups (or archives):
1. Full (also known as &quot;epoch&quot; or &quot;complete&quot;) — everything gets backed-up.
2. Incremental — backup everything that has been added or modified since the last backup of any type
(either incremental or full).
3. Differential — backup everything that has been added or modified since the last full backup. Differentials can be
assigned levels: level 0 is a full backup and level n is everything that has changed since the last level n−1 backup.
A system administrator must choose a backup strategy (a combination of types) based on several factors
(safety, required down time, cost, convenience, speed of backups and recovery, and so on).
* DRD is similar to Solaris Live Upgrade and AIX Alternate Root
* Old format:
# tctl –f /dev/rmt0 fsf 3 # tar -tvf/dev/rmt0
New format:
# tctl –f /dev/rmt0 fsf 3# restore -xvf /dev/rmt0 ./your/file/name
* System backup and create an /image.data file (generated by the mkszfile(1)).
** System backup and create an /image.data file with map files (generated by the mkszfile(1)).
*** System backup and create an /image.data file but exclude the files
in /home (create /etc/exclude.rootvg containing the line “/home”). It will
back up /home directory but not files it contains!
**** “-U” is for Universal Disk Format file system. It does not require the amount of
free space needed to create Rock Ridge format backups.
* Solaris ufsdump can, for example.
* DocId:UIUXKBRC00017271
In Ignite-UX revisions prior to c.6.2.x, both IA64 and PA-RISC systems could verify the contents of an
Ignite make_tape_recovery image by using the following method:
- Skip over the first image with mt command
# mt -f /dev/rmt/0mn rew
# mt -f /dev/rmt/0mn fsf 1
- Verify the contents of the tar image:
# tar -tvf /dev/rmt/0mn
After installing the new Ignite-UX verson c.6.2.x, this no longer works for make_tape_recovery
archives generated on Itanium systems. Tar errors out with a directory checksum error.
CONFIGURATION
Ignite-UX c.6.2.x HP-UX 11.x
RESOLUTION
According to the Ignite-UX release notes for version c.6.2.x
(http://www.docs.hp.com/en/IUX/docs/release_note.html) the following change has occurred:
- Itanium recovery tape format change
The format of Ignite-UX recovery tapes for Itanium systems has changed
to enable direct boot functionality. Direct boot may be supported
by some systems in the future. The new tape format includes standard
tape labels and a number of additional files needed for direct boot.
The two-step boot recovery process will continue to be supported and
legacy format recovery tapes will work using that process. No changes
have been made to PA-RISC recovery tape format or functionality.
The format of the header information on the tape has changed extensively. In order to get to the correct
spot on the tape and view the tar archive contents, you must now do the following:
# mt -f /dev/rmt/0mn rew
# mt -f /dev/rmt/0mn fsf 22
# tar -tvf /dev/rmt/0mn
* NFS Version 4 uses port 2049 only.
Courtesy of HP Education Training Materials in HE776 course
On a system with an LVM root, the LVM information is modified so that the booted volume group is always vg00.
The /dev/vg00 directory is removed from the clone and the /dev/drd00 directory is renamed /dev/vg00.
On a system with an LVM root, the LVM information is modified so that the booted volume group is always vg00.
The /dev/vg00 directory is removed from the clone and the /dev/drd00 directory is renamed /dev/vg00.
On a system with an LVM root, the LVM information is modified so that the booted volume group is always vg00.
The /dev/vg00 directory is removed from the clone and the /dev/drd00 directory is renamed /dev/vg00.
** To verify differences:
# diff /etc/passwd /var/opt/drd/mnts/sysimage_001/etc/passwd
For example, in Solaris, the whole comparison is automated:
# lustatus
Boot Environment Is Active Active Can Copy
Name Complete Now On Reboot Delete Status
-------------------------- -------- ------ --------- ------ ----------
d30 yes yes yes no -
BE2 no no no no ACTIVE
# lufslist BE1
# lufslist BE2
# lucompare BE2
SystemRescueCd. A Linux system on a bootable CD-ROM for repairing the server and data after a crash.
It also aims to provide an easy way to carry out admin tasks, such as creating and editing the partitions of the hard disk:
http://www.sysresccd.org/
Partition Image is a Linux/UNIX utility which saves partitions in many formats to an image file. The image file can
be compressed in the gzip/bzip2 formats to save disk space, and split into multiple files to be copied on removable floppies
(zip for example). Partitions can be saved across the network as well:
http://www.partimage.org/
mkCDrec makes a bootable disaster recovery image (CDrec.iso), including backups of the Linux system to the same
CD-ROM (or CD-RW) if space permits, or to a multi-volume CD-ROM set.
Otherwise, the backups can be stored on another local disk, NFS disk or (remote) tape.
After a disaster (disk crash or system intrusion) the system can be booted from the CD-ROM and one can restore the
complete system as it was (at the time mkCDrec was run) with the command /etc/recovery/start-restore.sh
Disk cloning (clone-dsk.sh script) allows one to restore a disk to another disk (the destination disk does not have to be of the
same size as it calculates the partition layout itself). A thrid script, restore-fs.sh, will restore only one filesystem to a partition
of your choice, and the user can choose with which filesystem the partition has to be formatted.
http://mkcdrec.ota.be/
Timo&apos;s Rescue CD:
http://rescuecd.sourceforge.net/
Linbox Rescue Server:
http://www.linbox.com/en/ppart.html
CloneZilla:
http://clonezilla.sourceforge.net/
Many other backup tools like Bacula, Amanda, Rsync, Rdiff-backup, and so on...
This site lists even more of them:
http://www.thefreecountry.com/utilities/backupandimage.shtml
Commercial and other possibilities:
SBAdminhttp://www.storix.com/linuxbackup/
Arkeia Softwarehttp://www.arkeia.com/
Acronis True Imagehttp://www.acronis.com/
R1Soft: http://www.r1soft/com/
LifeKeeperhttp://www.steeleye.com/
Ghost for Linuxhttp://freshmeat.net/projects/g4l/
PRIMECLUSTERNovell and Fujitsu
Altiris Deployment http://www.altiris.com/
And others.
Even a simple dd(1) with compression:
dd if=/dev/sdXX bs=1k conv=sync,noerror | gzip -c | dd of=PATH/filename.gz bs=1k
* Diskless Remote Boot in Linux: http://drbl.sourceforge.net/
* This command will create the ISO file clonezilla-live-myimage.iso which can then be burned to a CD or DVD.
** When you put the ISO in the CD/DVD and boot the CD/DVD, it will run in batch mode (-t), in English environment (-g en),
do NOT change the keyboard layout (-k NONE), and ocs-sr will run with parameters &quot;-g auto -b -c restoredisk myimg2 hda“
to restore image “myimg2&quot; into disk hda (-e &quot;-g auto -b -c restoredisk myimg2 hda&quot;). The above command will create
an ISO file &quot;clonezilla-live-myimg2.iso&quot;, and it will allow you to boot and run in batch mode.
*** Write customised script cust-osc and then create the ISO file for CD/DVD or
the zip file for USB flash drive.
* DUMP: bread: dev_seek error: error 0
DUMP: Warning - block 1082285240 is beyond the end of &apos;/dev/md/rdsk/d41&apos;
This is an unmistakable sign of a file system being
modified while backed up. ufsdump(1) does not support this.
For ufsdump(1), UFS must be unmounted or
remounted read-only, or at least be in a quiet state.
The modern alternative is to use fssnap(1) together
With ufsdump(1)…
** You can use the dry run mode (”-S” option) to determine the amount of space that
is needed before actually backing up the file system.
*** Ufsdump makes two passes when it backs up a file system. In the first pass, it scans
the raw device file for the file system and builds a table of directories and files in memory. Then,
it writes the table to the backup media. In the second pass, it goes through the inodes
in numerical order, reading the file contents and writing data to the backup media.
Courtesy of Sun Microsystems:
In some cases, you may be recovering a machine that has different peripherals than the master machine.
If you install the master system with the core, end user, developer, or entire software group, the master
system supports only the peripheral devices that are attached to the master machine at the time of backup.
With your disaster recovery plan, you should be able to recover a system that has different peripherals than
the master machine. For this reason, you should install the Entire Plus OEM software group on the master machine.
The flash archive created from the master machine with the Entire Plus OEM software group should work on any system
that has peripheral devices supported by the installed release of the Solaris O/S.
* /var/tmp/orbit-* are Unix domain sockets for interprocess communication between the JDS components.
Hence, you need to exclude them. For example:
# flar create –n sol10 –x /var/tmp/orbit-root /nfsdir/sol10-archive.flar
Diagram courtesy of Sun Microsystems
* You can save significant time with flarcreate by using the “-S” switch, stops it calculating the archive size.
* You can save significant time with flarcreate by using the “-S” switch, stops it calculating the archive size.
* You can save significant time with flarcreate by using the “-S” switch, stops it calculating the archive size.
* You can save significant time with flarcreate by using the “-S” switch, stops it calculating the archive size.
Errors generated:
Fssnap: write: error 28: No space left on device
Example:
Log in to the console of a zone running on serv1 and create a full flash
(this does not work properly with an image created from a global zone!):
zone1# flarcreate -n &quot;&quot; -S /var/tmp/zone1.flar (anywhere but /tmp)
If you don’t want to use the interactive prompt, you need to specify at least the -f, -k, -m, -t, and -s flags:
The -f flag forces btcreate to overwrite an existing file system.
The -k flag specifies the name of the kernel configuration file in /usr/sys/conf.
The -m flag allows you to specify a memory file system or a partition to use during the restore.
The -t flag specifies the device to back up to.
The -s flag allows you to specify a file that contains a list of file systems to back up.
NOTE: Bootable tapes are not supported on all systems and all drives - check the btcreate(1) man page or the Tru64 System Administration manual.
Common practice is to install CD-ROM O/S and restore a vdump session from tape when a system disk is lost.
If you don’t want to use the interactive prompt, you need to specify at least the -f, -k, -m, -t, and -s flags:
The -f flag forces btcreate to overwrite an existing file system.
The -k flag specifies the name of the kernel configuration file in /usr/sys/conf.
The -m flag allows you to specify a memory file system or a partition to use during the restore.
The -t flag specifies the device to back up to.
The -s flag allows you to specify a file that contains a list of file systems to back up.
NOTE: Bootable tapes are not supported on all systems and all drives - check the btcreate(1) man page or the Tru64 System Administration manual.
Common practice is to install CD-ROM O/S and restore a vdump session from tape when a system disk is lost.
* Courtesy of Eric van Dijken in ITRC Forums
* Data Protector Express and Data Protector are separate products that share
“Data Protector” name. There is no compatibility between the two products.