Q. Why?
A. Please go to the WHY page.
Q. What operating systems do I need to be running?
The burp server (where the backups are stored) needs to run on a Linux-style
operating system. Personally, I use Debian.
The burp client (from which backups are taken) can run on Linux or Windows.
Q. What are the main features? What works?
A. At the time of writing, the following features work:
- Server mode runs on Unix-based systems.
- Client mode runs on Windows and Unix-based systems.
- Windows Volume Shadow Copy Service (VSS) support (Windows XP R2 and newer).
- Windows 32bit and 64bit support.
- Delta differencing with librsync.
- Network backups.
- Backs up and restores files, directories, symlinks, hardlinks, fifos and
nodes.
- Storage and network compression using zlib.
- Ability to continue interrupted backups.
- SSL.
- Client side encryption - using this feature means that delta differencing
is pointless. If a file changes at all, the whole file will be transferred in
the next backup.
- Scheduling.
- Email backup success/failure notifications.
Q. What is on the roadmap? What doesn't work?
A. The following items are on my TODO list:
- Add ability to have pre/post client scripts.
- Investigate HFS / resource forks and Mac support.
- Add fifo read support.
- Add include/exclude wildcards.
- Think about having an ncurses style monitor for the server.
Probably best make some sort of generic interface that can have different
front-ends.
Q. What are the main dependencies?
A. They are:
- librsync
- openssl
- zlib (>=1.2.5 is recommended)
Q. Do I need to install cygwin on Windows?
No.
Q. How do I build the Windows client?
This is slightly complicated. You might want to go to the DOWNLOAD page and
get a prebuilt binary. If you really want to build it yourself, follow the
instructions in the README file in the src/win32 directory of the burp source.
Q. How do I build a static Linux binary?
Use the --enable-static and --disable-libtool configure script options.
Q. What licence does Burp use?
A. It is open source free software (where 'free' means both that you do not have
to pay for it, and that you have freedom to do what you want with it) released
under the AGPLv3 licence. Refer to the Burp
LICENCE file for more details.
Q. I get errors like "/0000/0000/1ACA.gz: File too large" in my log.
A. This is burp trying to open a backed up file on the burp server.
Older versions of zlib can have problems opening large files. Try upgrading
zlib to at least version 1.2.5.
Q. Are you going to do data deduplication on the server side?
No. There are file systems (like btrfs, zfs and btrfs) coming along that will
do data deduplication automatically. So applications won't need to do anything
particularly special just to get that feature, and anything similar that I
might implement now would probably be wasted effort.
Q. What is the best file system to use on the server?
So far, I have done all my testing on ext3, and it works for me. The main
concern is not to run out of file system inodes. You can configure the number
of inodes when you create the ext3 file system, but you cannot change the
number later without reformatting.
One of my test clients has over a million files. Because I had used
the ext3 optimisation option for large files on the storage device (having
previously been using it for bacula), I initially ran out of inodes.
Reformatting the file system to have more inodes has solved the problem.
To see how many inodes I have, I use the command 'df -i'
Here is a helpful 10-year old mailing list post about ext3 inodes:
http://markmail.org/message/ergkrg6thswonkgo
(ignore the recommendation for reiserfs - I believe that ext3 has moved on more
than reiserfs over the last 10 years!)
|
