Skip to main content

HardLink/Deduplication Backups with Python

Project description


Hardlink/Deduplication Backups with Python.

  • Backups should be saved as normal files in the filesystem:

    • accessible without any extra software or extra meta files
    • non-proprietary format
  • Create backups with versioning

    • every backup run creates a complete filesystem snapshot tree
    • every snapshot tree can be deleted, without affecting the other snapshots
  • Deduplication with hardlinks:

    • Store only changed files, all other via hardlinks
    • find duplicate files everywhere (even if renamed or moved files)
  • useable under Windows and Linux

Requirement: Python 3.4 or newer.

Please: try, fork and contribute! ;)

Build Status on
Build Status on
Coverage Status on
Requirements Status on


$ phlb backup ~/my/important/documents
...start backup, some time later...
$ phlb backup ~/my/important/documents

This will create deduplication backups like this:

  └── documents
      ├── 2016-01-07-085247
      │   ├── phlb_config.ini
      │   ├── spreadsheet.ods
      │   ├── brief.odt
      │   └── important_files.ext
      └── 2016-01-07-102310
          ├── phlb_config.ini
          ├── spreadsheet.ods
          ├── brief.odt
          └── important_files.ext



  1. install Python 3:
  2. Download the file boot_pyhardlinkbackup.cmd
  3. call boot_pyhardlinkbackup.cmd as admin (Right-click and use Run as administrator)

If everything works fine, you will get a venv here: %ProgramFiles%\PyHardLinkBackup

After the venv is created, call these scripts to finalize the setup:

  1. %ProgramFiles%\PyHardLinkBackup\phlb_edit_config.cmd - create a config .ini file
  2. %ProgramFiles%\PyHardLinkBackup\phlb_migrate_database.cmd - create database tables

To upgrade PyHardLinkBackup, call:

  1. %ProgramFiles%\PyHardLinkBackup\phlb_upgrade_PyHardLinkBackup.cmd

To start the Django webserver, call:

  1. %ProgramFiles%\PyHardLinkBackup\phlb_run_django_webserver.cmd


  1. Download the file
  2. call

Note: If you don’t use Python 3.5+, then you must install ‘scandir’, e.g.:

~ $ cd PyHardLinkBackup
~/PyHardLinkBackup $ source bin/activate
(PyHardLinkBackup) ~/PyHardLinkBackup $ pip install scandir

(You need the python3-dev package installed)

If everything works fine, you will get a venv here: ~\PyHardLinkBackup

After the venv is created, call these scripts to finalize the setup:

  • ~/PyHardLinkBackup/ - create a config .ini file
  • ~/PyHardLinkBackup/ - create database tables

To upgrade PyHardLinkBackup, call:

  • ~/PyHardLinkBackup/

To start the Django webserver, call:

  • ~/PyHardLinkBackup/

Starting a backup run

To start a backup run, use this helper script:

  • Windows batch: %ProgramFiles%\PyHardLinkBackup\PyHardLinkBackup_this_directory.cmd
  • Linux shell script: ~/PyHardLinkBackup/

Copy this file to a location that should be backed up and just call it to run a backup.

Verifying an existing backup

$ cd PyHardLinkBackup/
~/PyHardLinkBackup $ source bin/activate

(PyHardLinkBackup) ~/PyHardLinkBackup $ phlb verify --fast ~/PyHardLinkBackups/documents/2016-01-07-102310

With –fast the files’ contents will not be checked. If not given: The hashes from the files’ contents will be calculated and compared. Thus, every file must be completely read from filesystem, so it will take some time.

A verify run does:

  • Do all files in the backup exist?
  • Compare file sizes
  • Compare hashes from hash-file
  • Compare files’ modification timestamps
  • Calculate hashes from files’ contents and compare them (will be skipped if –fast used)


phlb will use a configuration file named: PyHardLinkBackup.ini

Search order is:

  1. current directory down to root
  2. user directory

E.g. if the current working directoy is /foo/bar/my_files/ then the search path will be:

  • /foo/bar/my_files/PyHardLinkBackup.ini
  • /foo/bar/PyHardLinkBackup.ini
  • /foo/PyHardLinkBackup.ini
  • /PyHardLinkBackup.ini
  • ~/PyHardLinkBackup.ini The user home directory under Windows/Linux

Create / edit default .ini

You can just open the editor with the user directory .ini file with:

(PyHardLinkBackup) ~/PyHardLinkBackup $ phlb config

The defaults are stored here: /phlb/config_defaults.ini

Excluding files/folders from backup:

There are two ways to exclude files/folders from your backup. Use the follow settings in your PyHardLinkBackup.ini

# Directory names that will be recursively excluded from backups (comma separated list!)
SKIP_DIRS= __pycache__, temp

# glob-style patterns to exclude files/folders from backups (used with Path.match(), Comma separated list!)
SKIP_PATTERNS= *.pyc, *.tmp, *.cache

The filesystem scan is divided into two steps: 1. Just scan the filesystem tree 2. Filter and load meta data for every directory item

The SKIP_DIRS is used in the first step. The SKIP_PATTERNS is used the the second step.

Upgrading PyHardLinkBackup

To upgrade to a new version just start this helper script:

Some notes

What is ‘phlb’ and ‘manage’ ?!?

phlb is a CLI.

manage is similar to a normal Django, but it always uses the PyHardLinkBackup settings.

Why in hell do you use Django?!?

  • Well, just because of the great database ORM and the Admin Site. ;)

How to go into the Django admin?

Just start:

  • Windows: phlb_run_django_webserver.cmd
  • Linux:

And then request ‘localhost’ (Note: –noreload is needed for Windows with venv!)

Running the unit tests

Just start: phlb_run_tests.cmd / or do this:

$ cd PyHardLinkBackup/
~/PyHardLinkBackup $ source bin/activate
(PyHardLinkBackup) ~/PyHardLinkBackup $ manage test

Using the CLI

$ cd PyHardLinkBackup/
~/PyHardLinkBackup $ source bin/activate
(PyHardLinkBackup) ~/PyHardLinkBackup $ phlb --help
Usage: phlb [OPTIONS] COMMAND [ARGS]...


  --version  Show the version and exit.
  --help     Show this message and exit.

  add     Scan all existing backup and add missing ones...
  backup  Start a Backup run
  config  Create/edit .ini config file
  helper  link helper files to given path
  verify  Verify a existing backup

Add missing backups to the database

phlb add can be used in different scenarios:

  • recreate the database
  • add a backup manually

phlb add does this:

  • scan the complete file tree under BACKUP_PATH (default: ~/PyHardLinkBackups)
  • recreate all hash files
  • add all files to database
  • deduplicate with hardlinks, if possible

So it’s possible to recreate the complete database:

  • delete the current .sqlite file
  • run phlb add to recreate

Another scenario is e.g.:

  • DSLR images are stored on a network drive.
  • You have already a copy of all files locally.
  • You would like to add the local copy to PyHardLinkBackup.

Do the following steps:

  • move the local files to a subdirectory below BACKUP_PATH
  • e.g.: ~/PyHardLinkBackups/pictures/2015-12-29-000015/
  • Note: the date format in the subdirectory name must match the SUB_DIR_FORMATTER in your config
  • run: phlb add

Now you can run phlb backup from your network drive to make a new, up-to-date backup.

Windows Development

Some notes about setting up a development environment on Windows: /dev/WindowsDevelopment.creole

Alternative solutions

See also:


  • dev - compare v0.11.0…master

    • TBC
  • 03.03.2019 - v0.11.0 - compare v0.10.1…v0.11.0

    • Update boot files
    • Use django v1.11.x
  • 09.09.2016 - v0.10.1 - compare v0.10.0…v0.10.1

  • 26.04.2016 - v0.10.0 - compare v0.9.1…v0.10.0

    • move under Windows the install location from %APPDATA%\PyHardLinkBackup to %ProgramFiles%\PyHardLinkBackup
    • to ‘migrate’: Just delete %APPDATA%\PyHardLinkBackup and install via boot_pyhardlinkbackup.cmd
  • 26.04.2016 - v0.9.1 - compare v0.9.0…v0.9.1

    • run migrate database in boot process
    • Add missing migrate scripts
  • 10.02.2016 - v0.9.0 - compare v0.8.0…v0.9.0

  • 04.02.2016 - v0.8.0 - compare v0.7.0…v0.8.0

    • New: add all missing backups to database: phlb add (s.above)
  • 03.02.2016 - v0.7.0 - compare v0.6.4…v0.7.0

    • New: verify a existing backup
    • IMPORTANT: run database migration is needed!
  • 01.02.2016 - v0.6.4 - compare v0.6.2…v0.6.4

    • Windows: Bugfix temp rename error, because of the Windows API limitation, see: #13
    • Linux: Bugfix scanner if symlink is broken
    • Display local variables on low level errors
  • 29.01.2016 - v0.6.3 - compare v0.6.2…v0.6.3

    • Less verbose and better information about SKIP_DIRS/SKIP_PATTERNS hits
  • 28.01.2016 - v0.6.2 - compare v0.6.1…v0.6.2

    • Handle unexpected errors and continue backup with the next file
    • Better handle interrupt key during execution
  • 28.01.2016 - v0.6.1 - compare v0.6.0…v0.6.1

    • Bugfix #13 by using a better temp rename routine
  • 28.01.2016 - v0.6.0 - compare v0.5.1…v0.6.0

    • New: faster backup by compare mtime/size only if old backup files exists
  • 27.01.2016 - v0.5.1 - compare v0.5.0…v0.5.1

    • IMPORTANT: run database migration is needed!
    • New .ini setting: LANGUAGE_CODE for change translation
    • mark if backup was finished compled
    • Display information of last backup run
    • Add more information into summary file
  • 27.01.2016 - v0.5.0 - compare v0.4.2…v0.5.0

    • refactory source tree scan. Split in two passed.

    • CHANGE SKIP_FILES in .ini config to: SKIP_PATTERNS

    • Backup from newest files to oldest files.

    • Fix #10:

      • New –name cli option (optional) to force a backup name.
      • Display error message if backup name can be found (e.g.: backup a root folder)
  • 22.01.2016 - v0.4.2 - compare v0.4.1…v0.4.2

  • 22.01.2016 - v0.4.1 - compare v0.4.0…v0.4.1

    • Skip files that can’t be read/write. (and try to backup the remaining files)
  • 21.01.2016 - v0.4.0 - compare v0.3.1…v0.4.0

    • Search for PyHardLinkBackup.ini file in every parent directory from the current working dir
    • increase default chunk size to 20MB
    • save summary and log file for every backup run
  • 15.01.2016 - v0.3.1 - compare v0.3.0…v0.3.1

    • fix unittest run under windows
  • 15.01.2016 - v0.3.0 - compare v0.2.0…v0.3.0

    • database migration needed
    • Add ‘no_link_source’ to database (e.g. Skip source, if 1024 links created under windows)
  • 14.01.2016 - v0.2.0 - compare v0.1.8…v0.2.0

    • good unittests coverage that covers the backup process
  • 08.01.2016 - v0.1.8 - compare v0.1.0alpha0…v0.1.8

    • install and runable under Windows
  • 06.01.2016 - v0.1.0alpha0 - d42a5c5

    • first Release on PyPi
  • 29.12.2015 - commit 2ce43

    • commit ‘Proof of concept’

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Filename, size & hash SHA256 hash help File type Python version Upload date
PyHardLinkBackup-0.11.0-py3.6.egg (65.7 kB) Copy SHA256 hash SHA256 Egg 3.6
PyHardLinkBackup-0.11.0-py3-none-any.whl (80.4 kB) Copy SHA256 hash SHA256 Wheel py3
PyHardLinkBackup-0.11.0.tar.gz (54.8 kB) Copy SHA256 hash SHA256 Source None

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN SignalFx SignalFx Supporter DigiCert DigiCert EV certificate StatusPage StatusPage Status page