Home
 |
An open source LVM backup utility for linux based systems. Barch conducts automatic and predefined volume structure recognition. Supports full/incremental backups. Based on duplicity. |
|---|
Contents:
Clone files to a temporary folder on target host, then run:
git clone git@github.com:sysboss/Barch.git
./INSTALL
Follow the installer instructions.
Barch v6.1 - LVM backup Solution
Copyright (c) 2015 Alexey Baikov <sysboss[@]mail.ru>
usage: /usr/local/bin/barch [options] FROM
Options:
-c|--cleanup Recovery mode
-o|--only Single logical volume to backup
-h|--help Help (this info)
-d|--daemon Run as a daemon
--graceful Graceful stop
--version Display version
--syntax Verify config file syntax
Debug Options:
-v|--verbose Log to stdout
--debug Debug mode (very verbose)
--dry-run Discover only mode
Once configuration complete, check barch.conf syntax with:
barch --syntax
Before the first execution it's highly recommended to create a test volume to prevent any possible damage!
That could be done by running Barch with --only option, to initiate single specified volume backup and --dry-run flag to skip the actual backup process, as shown:
barch --verbose --dry-run --only testLVname
Verbose mode will let you see the flow. --debug flag will show all the command executed.
In case of any failure or warning, see problems resolution section below.
Barch conducts LVM incremental backups by discovering entire hierarchy of each logical volume (such as partitions, file systems and even another LVM structure).
Before backup starts, Barch creates a snapshot of the volume and mount it to a temporary mount point. This method allows the backup tool to sync changes (increments) between the last backup and the current state.
Barch is written in Perl and depends on duplicity, rsync and kpartx. Please make sure this tools installed.
Default config file can be found at /etc/barch/barch.conf.
Adjust the following settings according your needs:
[global]
facility = deamon
lock_dir = /var/lock/barch
pidfile = barch.pid
work_dir = /var/cache/barch
report = /var/cache/barch/status
listen = 127.0.0.1
port = 3088
facility - syslog facility
report - report file contains information regarding each volume, such as backup status, last backup date etc.
Used by monitor.pl tool to check backups.
listen - bind address and port for the HTTP interface.
[default]
instance = name
# instance name used as prefix.
# hostname will be used by default.
# Handle DRBD
# check if volume is a drbd device (true/false)
check_drbd = false
# Handle LVM
# Backup volumes with internal LVM structure (true/false)
#backup_lvm = false
Notice
Internal volume group could have the same name as the local volume group (such as vg0).
Enable this option carefully to avoid naming conflict.
# skip filesystem types
skip_always = swap|luks
skip_always - Pipe separated file system types which will never be backed up.
# grace period for a new volumes
# since created (eg. 5M, 12H, 1D, 1W)
vol_grace = 24H
vol_grace - Wait period before backing up a new volume. 1M is a minimum.
# volume backup cycle
# eg. 20M, 1H, 1D, 1W
vol_cycle = 12H
vol_cycle - Defines the backup cycle for each volume
[snapshots]
# snapshot size
snap_size = 10G
# Snapshot size in MB/GB
# used to store changed since the snapshot was taken
max_snap = 90
# Abort backup if snapshot is full (in %)
This section is relevant for DRBD only. If check_drbd set to true.
[drbd]
# drbd states to backup Primary/Secondary/any
drbd_state = any
# connection states to backup Connected/Disconnected/any
drbd_conn = Connected
# device status UpToDate/Outdated/DUnknown/any
drbd_status = UpToDate
Notice params values are case sensitive. Use only allowed values.
# Remote cluster member
# with HTTP Interface enabled
#[cluster]
# member = x.x.x.2
# port = 3088
# # if cluster member unreachable
# # abort/continue
# on_error = abort
Uncomment this section to backup DRBD cluster in parallel from both nodes.
Remember to set source_mismatch to true in duplicity section below.
member - second node IP address
port - Barch instance port
on_error - Defines cluster behaviour if there is no connection between the nodes.
set continue - to do backups even if the second node is unreachable.
[duplicity]
# Perform a full backup every
full = 1W
# Number of retries to make on errors before giving up
retries = 3
# Delete all backup sets older than the given time
remove_older_than = 2W
# public GPG key ID
encrypt_key = ********
encrypt_pass = ****************************************
encrypt_key - is the gpg key 'sub' taken from gpg --list-keys
encrypt_pass - password of the gpg key
See GPG how-to: GPG Tutorial
# Destination backup folder
remote_snap_dir = rsync://barch@backupserv//usr/local/backup
# ftp://user[:password]@other.host[:port]/some_dir
# scp://user[:password]@other.host[:port]/some_dir
# file:///some_dir
# Change the volume size to number Mb. Default is 250Mb.
volsize = 250
# --allow-source-mismatch parameter (allow/deny)
source_mismatch = deny
# Allow this if you volume is a DRBD member
# to allow any host to backup it
maxtransfertime = 2H
maxtransfertime - is visibility feature only. Indicates maximum time for single chunk to be collected and transferred to backup server. Default is 2H.
[ssh]
# ssh credentials to remote storage
user = backup
server = x.x.x.x
path = /usr/local/backup
SSH credentials used to execute command and create sub-folders on the backup server.
[advanced]
# parallel execution
max_forks = 2
cpu_nice = CPU usage priority
Can range from -20 to 19, where lower numbers mean higher priority
io_nice = IO usage priority
Can range from 0 to 7, where lower numbers mean higher priority
custom = Custom specifications file location
Will be described later
This config file should contain custom specifications for more advanced filesystem types, such as ocfs2.
Example below, shows how custom.conf file can be used to define OCFS2 (Oracle Cluster File System 2) backup procedure.
[vg0.ocfs]
backup = true
fs_type = ocfs2
mount_opt = -o ro,localflocks
pre_mount = tunefs.ocfs2 -y -L ocfs_bsnap --cloned-volume /dev/vg0/ocfs_bsnap
post_mount = none
This config also used to exclude volumes from backup procedure by setting backup = false under relevant lv name.
backup - flag defines if LV should be backed up. If set to false, Barch will skip this LV.
No other flags required in this case.
fs_type - defines filesystem type (eg. ext4, ntfs, ocfs).
mount_opt - default Linux mount command parameters.
pre_mount - pre-mount script file path. Set 'none' if not required.
pre_backup - this script runs before snapshot creation. Optional.
Barch designed to run as a daemon (--daemon option).
Init script is also included.
Backup monitoring script can be found at /etc/barch/monitor.pl
Usage:
./monitoring.pl [OPTIONS ...]
Options:
-h|--help Help (this info)
-v|--verbose Verbose mode
-W|--warn Warning threshold (default: 18h)
-C|--crit Critical threshold (default: 1d)
By default, will monitor backup jobs status and verify that every volume was backed up in time.
Fell free to contact me if you have any questions. Answers will be published here.