Skip to content

Latest commit

 

History

History
353 lines (262 loc) · 9.72 KB

rear-presentation.adoc

File metadata and controls

353 lines (262 loc) · 9.72 KB
Raw

Relax-and-Recover: Disaster Recovery and System Migration solution (2015)

This presentation is written in AsciiDoc and can be converted to a LibreOffice/OpenOffice presentation using asciidoc-odf.

What is Relax-and-Recover ?

Tool that implements a DR workflow for Linux

Which basically means ?

  • Easy to deploy (set up and forget)

  • Makes recovery as easy as possible

  • Complete bare-metal disaster recovery

  • Integration for various Linux technologies

  • Integration with various back-up solutions

  • Modular framework written in Bash

  • Easy to extend to own needs

How does Relax-and-Recover work ?

Relax-and-Recover builds a rescue image from the existing installation

  • Restored system is an accurate copy

  • Restored system supports the hardware

  • Exotic setups (kernel, hardware) guaranteed to work

  • What worked before keeps on working

  • Recovery is fully automated

Backup software is integrated into rescue image

  • With a working configuration

Why is a backup not sufficient ?

Backup only covers storing data safely/securely

  • Often only the data is available

  • DR is a manual and time-consuming process

So manual disaster recovery risks:

  • unknown system storage and network configuration

  • incompatibilities between firmware/software/hardware

    • using newer tools with older data (or vice versa)

    • hardware is not supported

  • making manual mistakes (again and again)

  • working against the clock amidst chaos

  • frustration

What does Relax-and-Recover bring to the table ?

But the Relax-and-Recover rescue image takes care of the complete recovery process until the system runs:

  • enabling the network and remote access

  • using tools to recreate an accurate/working copy

  • recreating hardware and software RAID

  • partitioning and creating file systems

  • starting the backup restore process

  • restoring the bootloader

Relax-and-Recover functionality

  • Supported storage technologies

    • HP SmartArray, SW RAID, DRBD, LVM, multipath

    • ext2/ext3/ext4, xfs, jfs, vfat, btrfs, LUKS

  • Boot media or remote storage

    • OBDR tape, ISO, USB, eSATA, network booting

    • NFS, CIFS, rsync, HTTP, FTP, SFTP, …​

  • Back-up backends

    • IBM TSM, Micro Focus Data Protector, Symantec NetBackup, FDR/Upstream, Bacula, tar, rsync, Borg

How is Relax-and-Recover used ?

  • Creating boot media: rear mkrescue

    • Copies kernel, modules, minimal system

    • Saves storage layout, network configuration

    • Write boot media

  • Restore system

    • Boot Relax-and-Recover media

    • Optionally, modify storage/network config

    • Run: rear recover

    • Relax-and-Recover parses layout then creates diskrestore script

    • Relax-and-Recover initiates restore or prompts for restore

Relax-and-Recover on a running system

/-------------+       /-----------+       /-----------+
| Storage     |       | Collect   |       | Integrate |
| layout      | ----> | new       | ----> | required  |
| change      |       | storage   |       | backup    |
| detected    |       | layout    |       | strategy  |
+-------------/       +-----------/       +-----------/

                                                |
                                                V

/-------------+       /-----------+       /-----------+
| Off-site    |       | Push      |       | Create    |
| storage     | <---- | recovery  | <---- | new       |
| together    |       | image     |       | recovery  |
| with backup |       | centrally |       | image     |
+-------------/       +-----------/       +-----------/

Nifty features to help relax

  • Local GRUB integration (password protected)

  • Serial console support (think: disaster)

  • History-stuffing during recovery

  • Network and SSH key integration

  • Layout code guides you through recovery

    • Menu’s and command-line in one session

    • Provides original storage info

  • Beep, UID led and USB suspend integration

  • Syslinux management

  • Log-file on recovery media

Use case: Belgian Federal Police /1

  • Requirements:

    • About 200 sites with each a set of Linux servers

    • Each server comes with a tape-drive

    • Single bootable tape to:

      • Restore complete system

      • Restore from back-up

    • Support for various technologies

      • HP SmartArray, SW RAID, DRBD, LVM

      • OBDR, Bacula tape support

      • RHEL 4, RHEL5 and RHEL6 support

    • End-user documentation in 3 languages

Use case: Belgian Federal Police /2

  • Solution:

    • All requirements implement but…​

    • New systems didn’t support bootable tape (OBDR)

    • USB is much more flexible than tape

      • Can store multiple rescue images

      • Can store rescue images of multiple servers

      • Easier workflow (udev): insert, wait, pull (2 min max)

      • Cheaper

      • Implementation is more reliable

    • Bonus implementation of flexible layout

    • Support migration scenarios

Use case: Belgian Federal Police /3

  • Relax-and-Recover config for USB rescue media:

BACKUP=BACULA
OUTPUT=USB
USB_DEVICE=/dev/disk/by-label/REAR-000

  • Relax-and-Recover config for USB backup media:

BACKUP=NETFS
OUTPUT=USB
USB_DEVICE=/dev/disk/by-label/REAR-000
ONLY_INCLUDE_VG=( vg00 )
EXCLUDE_MOUNTPOINTS=( /data )

Use case: Belgian Federal Police /4

  • Relax-and-Recover config for OBDR rescue tapes:

BACKUP=BACULA
OUTPUT=OBDR
BEXTRACT_DEVICE=Ultrium-1
BEXTRACT_VOLUME=VOL-*

  • Relax-and-Recover config for OBDR backup tapes:

BACKUP=NETFS
OUTPUT=OBDR
TAPE_DEVICE=/dev/nst0

Use case: Centralized images /1

  • Requirements:

    • Remote rescue images

    • Removable media for off-site storage

    • Easy restore of physical hosts and guests

  • Solution:

    • Cron creates image when Relax-and-Recover detects change

    • Images pushed through HTTP to PXE server/host

    • Stored on USB disks, rotated every week

Use case: Centralized images /2

  • Label USB disk(s) and mount

  • Configure Apache to allow HTTP PUT to USB disk

  • Relax-and-Recover config in /etc/rear/local.conf:

BACKUP=BACULA
OUTPUT=ISO
ISO_URL=http://server:port/path/

  • Relax-and-Recover cron-job at /etc/cron.d/rear:

30 0 1 * * root /usr/sbin/rear mkrescue
30 1 * * * root /usr/sbin/rear checklayout || /usr/sbin/rear mkrescue

Relax-and-Recover command line /1

[root@moria rear]# rear help

Usage: $PROGRAM [-h|--help] [-V|--version] [-dsSv] [-D|--debugscripts SET] [-c DIR] [-r KERNEL] [--] COMMAND [ARGS...]
-snip-
List of commands:
 checklayout     check if the disk layout has changed
 dump            dump configuration and system information
 format          format and label media for use with rear
 mkbackup        create rescue media and backup system
 mkbackuponly    backup system without creating rescue media
 mkrescue        create rescue media only
 recover         recover the system; valid during rescue
 validate        submit validation information

Use 'rear -v help' for more advanced commands.

Relax-and-Recover command line /2

[root@moria rear]# rear help

Usage: $PROGRAM [-h|--help] [-V|--version] [-dsSv] [-D|--debugscripts SET] [-c DIR] [-r KERNEL] [--] COMMAND [ARGS...]

Available options:
 -h --help           usage information
 -c DIR              alternative config directory; instead of /etc/rear
 -d                  debug mode; run many commands verbosely with debug messages in log file (also sets -v)
 -D                  debugscript mode; log executed commands via 'set -x' (also sets -v and -d)
 --debugscripts SET  same as -d -v -D but debugscript mode with 'set -SET'
 -r KERNEL           kernel version to use; currently '4.12.13-123.45.67-default'
 -s                  simulation mode; show what scripts are run (without executing them)
 -S                  step-by-step mode; acknowledge each script individually
 -v                  verbose mode; show messages what Relax-and-Recover is doing on the terminal
 -V --version        version information

-snip-

Hacking on Relax-and-Recover

  • It is Bash !

*Join the mailinglist

  • Understand modular framework and workflows

    • Use: rear -s <workflow>

  • Logging

    • Logfile in: /var/log/rear/rear-<hostname>.log

  • Debugging

    • Verbose: rear -v

    • Debug: rear -d

    • Tracing: rear -D

Project future

  • Functionality

    • Improved rsync support (like rsnapshot or rbme)

    • More back-up backend integration

    • PXE integration

  • Development

    • Re-organize code base

    • Release management needs a process

    • Website and documentation not up-to-date

    • Change of development tools ?

Development team

Consists of:

  • Schlomo Schapiro (original author)

  • Gratien D’haese (original author)

  • Jeroen Hoekx (new contributor)

  • Dag Wieers (new contributor)

and various other contributors

Development at Github

Thank you for listening

Any questions, ideas, pull-requests ?

Live USB demo

  • Demo USB/udev integration (check udev config !)

    • Prepare USB stick: rear format /dev/sdb

    • Re-insert USB stick and wait until light goes out

  • Demo restore procedure (disable udev rule !)

    • Boot from KVM virtual machine

    • Re-insert USB stick and start virt-manager

    • Create VM with 4GB disk and boot VM

    • Show bash history and perform: rear recover

    • Show menu system and modify sizes