XenServer/XCP-ng VM Backup
OnyxBackupVM
Copyright (c) 2018 OnyxFire, Inc.
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
- The OnyxBackupVM tool is run from a XenServer host and utilizes the native
xe vm-export
andxe vdi-export
commands to backup both Linux and Windows VMs. - The backup is run after a respective vm-snapshot or vdi-snapshot occurs, which allows for the backup to execute while the VM is up and running.
- During the backup of specified VMs, this tool collects additional VM metadata using XAPI. This additional information can be useful during VM restore situations and is stored in ".meta" files.
- Typically, OnyxBackupVM is implemented through scheduled crontab entries or can be run manually on a XenServer ssh session. It is important to keep in mind that the backup process does use critical dom0 resources, so running a backup during heavy workloads should be avoided (especially if used with
compress
option). - The SRs where one or more VDIs are located require sufficient free space to hold a complete snapshot of a VM. The temporary snapshots that are created during the backup process are deleted after the backup has completed.
- OnyxBackupVM will require lots of file storage; setup a storage server with an exported VM backup share.
- Frequently NFS is used for the storage server, with many installation and configuration sources available on the web.
- An optional SMB/CIFS mode can be enabled via the
share_type
option in the config file.
- For all of the XenServers in a given pool, mount the new share at your desired filesystem location.
- NOTE: Make sure to add the new mount information to the
/etc/fstab
file to ensure it is remounted on a reboot of the host.
- NOTE: Make sure to add the new mount information to the
- Download and extract the latest release to your desired execution location, such as
/mnt/onyxbackup
- Generally you would extract to and run OnyxBackupVM from the mounted backup share so that the same version, backups, logs, and configuration are visible to all XenServer hosts, though this is not a requirement
<OnyxBackupVM path>/etc
- Contains
onyxbackup.example
(example onyxbackup.cfg file that is heavily commented) - Contains
logging.example
(example of logging.json with default logging configuration) - Location for optional configuration file
onyxbackup.cfg
for overriding default configuration - Location for optional
logging.json
for overriding default logging settings
- Contains
<OnyxBackupVM path>/logs
- Contains OnyxBackupVM log files
OnyxBackupVM.log
anddebug.log
based upon default logging configuration
- Contains OnyxBackupVM log files
<OnyxBackupVM path>/exports
- Contains all the VM/VDI backups
- Can be independently configured to be located wherever you desire using the
backup_dir
option
- Inspect and customize certain options in the
<OnyxBackupVM path>/etc/onyxbackup.cfg
,/etc/onyxbackup.cfg
, and/or~/onyxbackup.cfg
as desired.- The configuration files are read in the following order with a "last match wins" convention
<OnyxBackupVM path>/etc/onyxbackup.cfg
/etc/onyxbackup.cfg
~/onyxbackup.cfg
- Optional config file using
-c <file>
or--config <file>
command-line option
- The configuration files are read in the following order with a "last match wins" convention
- Initially use the
--preview
command-line option to confirm the resulting configuration from files and command-line flags before running an actual backup. - Set up a crontab entry or method for executing backups on a schedule
- You can run OnyxBackupVM from any host in the pool with all available options, not just the master; running it on a host with little to no VMs on it may be optimal since dom0 resources are used for backups.
onyxbackup-vm.py [-h] [-v] [-c FILE] [-d PATH] [-p] [-H] [-l LEVEL] [-C] [-F FORMAT] [--preview] [-e STRING]
[-E STRING] [-x STRING]
optional arguments:
-h, --help
show this help message and exit
-v, --version
show program's version number and exit
-c FILE, --config FILE
Config file for runtime overrides
-d PATH, --backup-dir PATH
Backups directory (Default: <OnyxBackupVM path>/exports
)
-p, --pool-backup
Backup Pool DB
-H, --host-backup
Backup Hosts in Pool (dom0)
-l LEVEL, --log-level LEVEL
Log Level (Default: info
)
-C, --compress
Compress on export (vm-exports only)
-F FORMAT, --format FORMAT
VDI export format (vdi-exports only, Default: raw
)
--preview
Preview resulting config and exit
-e STRING, --vm-export STRING
VM name or Regex for vm-export (Default: .*
) NOTE: Specify multiple times for multiple values
-E STRING, --vdi-export STRING
VM name or Regex for vdi-export (Default: None
) NOTE: Specify multiple times for multiple values
-x STRING, --exclude STRING
VM name or Regex to exclude (Default: None
) NOTE: Specify multiple times for multiple values
# Backup all VMs in the pool
./onyxbackup-vm.py
# Backup a single VM by name (case sensitive)
./onyxbackup-vm.py -e 'DEV-MYSQL'
# Backup a single VM by name with spaces in name
./onyxbackup-vm.py -e 'DEV MYSQL'
# Backup VMs by regex which matches more than one VM
./onyxbackup-vm.py -e 'DEV-MY.*'
# Backup VM by name and keep last 2 backups (overrides max_backups)
./onyxbackup-vm.py -e 'DEV-MYSQL:2'
# Export just root disk (xvda) for a single VM by name
./onyxbackup-vm.py -E 'DEV-MYSQL'
# Export just xvdb disk from a single VM by name without overriding number of backups to keep
./onyxbackup-vm.py -E 'DEV-MYSQL:-1:xvdb'
# Export 2 disks of a VM by name and keep last 7 backups
./onyxbackup-vm.py -E 'DEV-MYSQL:7:xvda;xvdb'
# A mix of the options to show some typical selections if not using config file to specify VMs
./onyxbackup-vm.py -e 'PRD-.*' -e 'MYSQL123' -E 'APPSERVER01:8:xvda;xvdc' -e 'DEV-.*' -x 'DEV-SHORTtest' -p -H
For the handling of wildcard VMs, OnyxBackupVM incorporates the native python regex library of regular expressions. WARNING: The syntax is slightly different from what is processed with the Linux family of grep commands. In python, a string followed by *
causes the resulting regular expression to match zero or more repetitions of the preceding regular expression, as many repetitions as are possible. For example, ab*
will match "a", "ab", or "a" followed by any number of "b" characters. Therefore, if you had PRD-a*, this will match PRD-a, PRD-aa, PRD-aaaSomething, as well as PRD- but also PRD-Test, PRD-123 and anything else starting with PRD- since zero occurences of the "a" after the "-" are matched. To avoid this, PRD-a.* should be used, instead, indicating in this case PRD-a followed by any single character (".") zero or more("*") times.
Note that the current implementation uses the re.match function which by design is expected to be front-anchored. You can explicitly preface a search string with the "^" operator, if desired, but it is already implied and the results will be the same.
A VM name is considered to be a simple (non-wildcard) name, provided that it contains only combinations of any of the following: letters (upper as well as lower case), numerals 0 through 9, the space character, hyphens (dashes), and underscore characters. These will not be handled using regex operations!
Configuration of backups follows the below order of precendence and lower precedence matches will not match the same VM already matched by a higher precedence match:
- Excludes
- VDI Exports (just disk)
- VM Exports (full backup)
Exclude simple host names (non-wildcard instances)
excludes = DEV-phishing-test,PRD-PRINTER1
Exclude using end-of-string - handled as a regex expression
excludes = PRD-PRINT$
Exclude PRD-a as well as PRD-a followed by anything (note the use of .*
instead of *
)
excludes = PRD-a.*
Exclude DEV-V followed by anything and ending in TEMP (end-of-string)
excludes = DEV-V.*TEMP$
Backup just the root VDI of DEV-web3
vdi_exports = DEV-web3
Same as above, except retain seven copies
vdi_exports = DEV-web3:7
Same again except use default max backups and backup first two disks (note semi-colon separator)
vdi_exports = DEV-web3:-1:xvda;xvdb
Backup just the root VDI for anything starting with DEV followed by anything, followed
by either the string "TEST" or the string "test" followed by anything else and also backup a
specific VM
vdi_exports = DEV-.*(TEST|test).*,DEV-EUWWW01
Same as above, except using the string "Test" or "test"
vdi_exports = DEV-.*[Tt]est.*
Backup VMs ending with the string SAVE and keep five backups
vm_exports = .*SAVE$:5
Backup VMs starting with PRD- and ending with print4 or print5 and save five backups and
backup another specific VM and keep only 2 backups
vm_exports = PRD-.*print[4-5]$:5,DEV-USWWW05:2
Backup VMs starting with PRD- and ending with print6, print7 or print8 and save ten versions
vm_exports = PRD-.*print[6-8]$:10
Backup VMs starting with PRD- and ending with print followed by any number(s) and ending
in that same string of numbers and save eight versions
vm_exports = PRD-.*print[\d{1,}]$:8
Note that there are numerous combinations that may possibly conflict with each other or potentially overlap. It is strongly encouraged to use the --preview
option to review the configuration output before putting into service. Also note that excludes
, vdi_exports
, and vm_exports
options should only be specified once each in the configuration files with multiple values being comma-separated.
Run backup once a week
10 0 * * 6 <OnyxBackupVM path>/onyxbackup-vm.py >/dev/null 2>&1
Run backup once a week and only log warnings and above
10 0 * * 6 <OnyxBackupVM path>/onyxbackup-vm.py -l warning >/dev/null 2>&1
Run backup of all VMs nightly and backup pool metadata and hosts weekly
10 0 * * * <OnyxBackupVM path>/onyxbackup-vm.py >/dev/null 2>&1
10 0 * * 6 <OnyxBackupVM path>/onyxbackup-vm.py -x '.*' -p -H >/dev/null 2>&1
The number of VM backups saved is based upon the configured max_backups value. For example, if max_backups=3 and the fourth successful backup completes, the oldest backup will be deleted. The vm_exports and vdi_exports each have their associated process list where each entry is of the form vm-name/regex:max_backups. The :max_backups is optional, and, if specified, is the maximum number of backups to maintain for this vm-name. Otherwise, the global max_backups is in effect for the given vm-name. At the completion of every successful VM vm-export/vdi-export operation, the oldest backup(s) are deleted using the in effect vm-name:max_backups value. If you want to specify specific disks to backup during a vdi-export, you must specify the max_backups field; if you do not want to deviate from the configured setting just use -1 as the value (i.e. VMNAME:-1:xvdb;xvdc
).
WARNING: Each VDI backed up using vdi-export counts as a backup, even if for the same VM, so keep this in mind if you specify multiple disks for a VM (i.e. MYVM03:2:xvda;xvdb
will only keep one backup of each disk since together they total 2 backups).
The following VM selection operations apply to the vm-export/vdi-export configuration (both command-line and config file selections): (1) Remove any matched VMs from excludes (both simple and regex-based) from the available list of VMs in the pool, (2) load each matched VM in vdi_exports into the config for vdi-export, then finally (3) load each matched VM from vm_exports into the config for full export ignoring any VMs already marked for vdi-export. By using the --preview
option the scope of the given OnyxBackupVM run is clearly output and is a good way to test.
For any individual OnyxBackupVM run, a VM found in the vdi-export list will take precedence over a matching entry in the vm-export list. The convention is that a VM is either backed up with a vm-export or a vdi-export, but not both. If at some point in time a VM grows in number of /dev/xvdX disks where it is required to switch from vm-export to vdi-export, then the same %BACKUP_DIR%/vm-name structure continues. Since, in this case, the backups are ordered by date with a mix of vdi-export and older vm-export backups, eventually the vm-export backups will be deleted based upon configured max_backups
.
The VM backup directory has this format %BACKUP_DIR%/vm-name/ and each VM backup directory contains the vm backup files plus backup metadata files from each backup.
The vm backup file has one of four possible formats, (1) backup_[date]-[time].xva which is created from a vm-export, (2) backup_[date]-[time].xva.gz created from a vm-export with compress
option, (3) backup_[date]-[time].raw which is created from vdi-export in raw format, or (4) backup_[date]-[time].vhd which is created from vdi-export in vhd format.
For each backup, there is a dump of selected XenServer VM metadata in a backup_[date]-[time].meta file. This information can be useful in certain recovery situations:
- VM
- name_label, name_description, memory_dynamic_max, VCPUs_max, VCPUs_at_startup, os_version, orig_uuid
- DISK (for each attached disk)
- device, userdevice, bootable, mode, type, unpluggable, empty, orig_uuid
- VDI
- name_label, name_description, virtual_size, type, sharable, read_only, orig_uuid, orig_sr_uuid
- VIFs (for each attached VIF)
- device, network_name_label, MTU, MAC, other_config, orig_uuid
Use the xe vm-import
command. See xe help vm-import
for parameter options. In particular, attention should be paid to the "preserve" option, which if specified as preserve=true
will re-create as many of the original settings as possible, such as the associated VM UUID values along with the network and MAC addresses.
Use the xe vdi-import
command. See xe help vdi-import
for parameter options. The current Citrix documentation is lacking and the best vdi-import examples can be found at http://wiki.xensource.com/wiki/Disk_import/export_APIs
Consult the Citrix XenServer Administrator's Guide chapter 8 and review sections that discuss the xe pool-restore-database
command.
- If
pool_backup
option has been specified then a %BACKUP_DIR%/POOL_DB/metadata_[date]-[time].db file will be created.
Use the xe host-restore
command. See xe help host-restore
for parameter options.
- If
host_backup
option has been specified then a %BACKUP_DIR%/HOST_[hostname]/host_[date]-[time].xbk file will be created for each host in the pool.
This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
This program was inspired by the great work of Northern Arizona University IT Department on NAUBackup. Please note that this is a completely rewritten program from the ground up and, as such, it has and will continue to evolve into something new.