Welcome! » Log In » Create A New Profile

link_dest and error-handling.

Posted by Anonymous 
link_dest and error-handling.
November 27, 2007 12:25AM
The recent discussion of re-using hourly.5 rather than cp -al on
hourly.0 got me thinking about whether I want to enable link_dest. My
backup server is running Ubuntu 6.04, and thus rsnapshot 1.2.1, but
I've also looked at 1.3.0's source code, and AFAICT while link_dest
correctly handles detected errors, I can't figure out how it handles
things like power failure (or rogue sysadmin :-).

What I want to happen is something like:

rsync ... --link-dest=.../hourly.0/myhost/ myhost:/ hourly.tmp/myhost/

cycling through all of my backup points, and then ending with rotating
hourly.tmp into hourly.? as you'd expect. If rsnapshot detects an
error during the backup, then one could use the cp -al trick on that
backup point, while continuing to finish the others (I believe this is
what the current link_dest rollback code does). If rsnapshot hourly
starts up and sees an hourly.tmp directory, that means that things are
foobar, so it deletes that directory before proceeding as normal.

I'm not asking someone to implement this - I just want to know if I'm
reading this correctly?

[I _think_ that combining sync_first and link_dest can sort-of kind-of
give you the behaviour I describe, but it seems like it's going the
long way around. Also, it was recently posted that there's a bug with
that combination, and as far as I understand the bug, it wouldn't
occur if you were building an hourly.tmp target from the ground up.]


This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
rsnapshot-discuss mailing list
rsnapshot-discuss < at > lists.sourceforge.net
Sorry, only registered users may post in this forum.

Click here to login