My system assumes two things: there is a repository folder and a backup
folder; and the scripts have complete control over the backup folder. My
folder layout looks something like:
E:
|
+- svnback
|
+- svnrepo
|
+- repo1
|
+- folder
|
+- repo2
|
+- repo3
The backup scripts go in the svnback folder and I have a scheduled event
to run them once a day. The backup scans the repository folder and
mimics its folder structure. Then determines the locations of any
repositories. A check is then done between the backup folder and the
repository to see the last backup revision. If the last checked in
revision is greater than the last backup revision, an incremental backup
is done and compressed (using gzip).
The restore is even simpler. It creates a new folder (svnrestore), and
starts recreating the entire folder structure and loading all database
info.
The batch files are present so I have a one-click solution to backing up
and restoring (they esentially have the paths I use). The only thing I
don't like about this, is that new repositories must have their first
complete dump done manually. This isn't that big a deal, all I did was
write that up as part of the process for creating a new repository.
If you have something signifigantly different, I would love to see it.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Wed Jun 29 18:12:57 2005