https://github.com/xor-gate/svn-backup2
Frugal and efficient Subversion backup solution
https://github.com/xor-gate/svn-backup2
backup backup-cli backup-script backup-solution subversion subversion-backup
Last synced: about 2 months ago
JSON representation
Frugal and efficient Subversion backup solution
- Host: GitHub
- URL: https://github.com/xor-gate/svn-backup2
- Owner: xor-gate
- License: mit
- Created: 2024-01-25T08:25:09.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2024-07-01T11:44:48.000Z (10 months ago)
- Last Synced: 2025-01-13T16:17:21.257Z (3 months ago)
- Topics: backup, backup-cli, backup-script, backup-solution, subversion, subversion-backup
- Language: Shell
- Homepage:
- Size: 73.2 KB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# svn-backup2
The `svn-backup2` application is a frugal Subversion backup solution inspired by [adamonduty/svn-backup](https://github.com/adamonduty/svn-backup).
Orignally written and tested for Debian 12 in Bash but could be adapted to other systems.
At first run it generates a full-backup and runs `svnadmin verify`.
This script is designed to be run as root. But it is possible when correct permissions are given to run as non-root user.
## Features and limitations
### Single file mode
It creates and updates a single file per repository and consists of multiple gzip compressed archives. It is sha256 checksumed to
detect bit-rot before appending a new incremental backup. This is possible because `svnadmin load` is able to load from a stream of multiple dumps.
This is not rsync efficient.
* Set `SVNBACKUP2_CFG_USE_SINGLE_FILE=1` (and `SVNBACKUP2_CFG_BACKUP_NR_DATASETS=0`)
* Only supports one subversion root path which can contain multiple repositories
* Single repository backup file for simplicity of restoration
* Can only restore the whole backup file into an empty repository
* `${SVNBACKUP2_CFG_BACKUP_PATH}/.svnbackup2`: Backup file (multiple concatinated gzip files) (configurable)
### Dataset mode
In dataset mode a full backup is written and per incremental run a new file is created. This is rsync efficient.
* Set `SVNBACKUP2_CFG_BACKUP_NR_DATASETS=1` or higher (and `SVNBACKUP2_CFG_USE_SINGLE_FILE=0`)
* `${SVNBACKUP2_CFG_BACKUP_PATH}//dataset_/_full.svnbackup2`: Full backup in dataset n
* `${SVNBACKUP2_CFG_BACKUP_PATH}//dataset_/_incr_rev-rev.svnbackup2`: Incremental backup in dataset n
## Installation
Install dependencies on Debian Linux 12:
`apt install subversion curl`
Download and install version `0.2.2` to `/usr/local/bin/svn-backup2` in a oneliner with curl:
```
FILE=/usr/local/bin/svn-backup2 &&
curl -o ${FILE} https://raw.githubusercontent.com/xor-gate/svn-backup2/v0.2.2/svn-backup2 &&
chmod -v 755 ${FILE} &&
chown -v root:root ${FILE}
```
## Paths
* `/etc/svn-backup2.conf`: Configuration file
* `/var/log/svn-backup2.log`: Logfile
* `/srv/svn`: Subversion repositories root path (configurable)
* `/var/tmp/svn-backup2/.state`: State of a single repository backup
## Configuration file (`/etc/svn-backup2.conf`)
The file is a set of shell environment variables. It will be created on first run
with the following defaults:
```
SVNBACKUP2_CFG_VERSION=1
SVNBACKUP2_CFG_REPOSITORIES_PATH="/srv/svn"
SVNBACKUP2_CFG_BACKUP_PATH="/srv/svn-backup2"
SVNBACKUP2_CFG_USE_SINGLE_FILE=1
SVNBACKUP2_CFG_BACKUP_NR_DATASETS=0
```
## Repository backup state file
A single repository backup state file is a dump of `SVNBACKUP2_REPO_STATE_*` environment variables
Example:
```
SVNBACKUP2_REPO_STATE_CHECKSUM=a911966efe3b069c970ae5511e4a3e6c169bf993e9a3443d841d5d9002447388
SVNBACKUP2_REPO_STATE_LAST_BACKUP_DURATION=1
SVNBACKUP2_REPO_STATE_LAST_BACKUP_END_TIME=1706279396
SVNBACKUP2_REPO_STATE_LAST_BACKUP_START_TIME=1706279396
SVNBACKUP2_REPO_STATE_YOUNGEST=0
```
## Usage
```
Usage: svn-backup2 [operation]
none All repositories are checked and backuped
version Print the version
full [ | all] Force full backup on given repository or "all"
```
## Cron in dataset mode
We use cron to cycle between the different datasets. As an example we generate a full
backup once a month. And the other days an incremental:
```
5 2 1 * * /usr/local/bin/svn-backup2 full all
5 2 2-31 * * /usr/local/bin/svn-backup2
```
## Restoration of single file
svn-backup2 generates simple subversion dumpfiles that can be stream loaded into
`svnadmin load`.
Step 1: Create an empty repository
```
svnadmin create /srv/svn/repository_1
```
Step 2: Uncompress the dumpfile as stream and load into svnadmin
```
zcat /srv/svn-backup2/repository_1.svnbackup2 | svnadmin load /srv/svn/repository_1
```
Step 3: There is no step 3!
A simple shell command can load all repositories at once:
```
find . -type f -name "*.svnbackup2" | while read i; do repository_name=`basename "$i" .svnbackup2`; svnadmin create "$repository_name" && zcat "$i" | svnadmin load "$repository_name"; done
```
## Restoration of dataset
With `ls` you can sort on creation date and load it with `svnadmin`
```
$ svnadmin create /tmp/testrepo
$ for backup in `ls -1 -r -t --time=creation /srv/svn-backup2/testrepo2/dataset_1/*`; do zcat $backup | svnadmin load /tmp/testrepo; done
```
# License
svn-backup2 is copyright 2024 by Jerry Jacobs and distributed under the terms of the MIT License. See the [LICENSE](LICENSE) file for further information.
## Alternatives
* [adamonduty/svn-backup](https://github.com/adamonduty/svn-backup) written in Ruby (defunc on Debian 12)
* [loresoft/SvnTools](https://github.com/loresoft/SvnTools) written in C#
* [ghstwhl/SVNBackup](https://github.com/ghstwhl/SVNBackup) written in Perl
* [gcraig/SVNBackup](https://github.com/gcraig/SVNBackup) written in Python
* [subversion-tools](https://packages.debian.org/sid/subversion-tools) multiple tools written in C/Perl
* [avalax/svnbackup](https://github.com/avalax/svnbackup) shell script using svn-hot-backup