Q1: Has anyone used rsnapshot on a compressed filesystem?
After doing some research, it looks like "FuseCompress" is the only
read-write, compressed filesystem for Linux that is mature and in active
development. But "mature" may be too strong a word; there are no
packages available, only a git checkout.
I've played with it, and it seems pretty stable. It appears to
support hard links; this means it should be usable with rsnapshot. (The
advantage over BackupPC is that you can share your backup snapshots
using Samba, Netatalk, WebDAV, etc.)
I'll play with it some more. Unfortunately, FuseCompress uses a
proprietary (but open) format for the compressed files. I was hoping it
would transparently convert any "file.txt" into "file.txt.gz", so that
the backup store was useful even without FuseCompress installed. But
instead, you need to re-mount the FuseCompress store to work with it.
--Derek
On 10/11/2009 09:23 PM, John Locke wrote:
Hey, Derek,
Haven't done either of the things you're asking for, but did want to
point out that BackupPC has some shell commands available for extracting
files from the backups--you don't need to use the web interface to do
the restore.
However, it is still slow--it took hours to extract about 30GB the one
time we had to do a total restore.
Otherwise, we've had really great experiences with BackupPC--restoring
individual files from it seems to be something we do at least monthly.
It's a great system, and very convenient...
Cheers,
John
-------- Original Message --------
Subject: Re: [SLL] rsnapshot
From: Derek Simkowiak <[email protected]>
To: Mark Foster <[email protected]>
Cc: [email protected]
Date: Sun 11 Oct 2009 01:06:09 PM PDT
I'm using rsnapshot on all my production servers. I love it. For
one client, I simply exported the rsnapshot dir with Samba as
read-only, and they love it.
BackupPC seems pretty popular. It has a nice web interface. It's
also based on rsync, and it also only stores the deltas (using
hardlinks). But it stores the files in an open-yet-proprietary
format. You are forced to use the web interface to retrieve files,
which is not always handy in an emergency situation.
On the plus side, BackupPC compresses the stored files, and it also
caches the MD5s of the files (which is a big deal for network-based
backups). I wish rsnapshot had those features.
I'm curious if anyone has done these setups:
Q1: Has anyone used rsnapshot on a compressed filesystem? Something
like compFUSEd, FuseCompress, or fuse-zip? If one of those fuse-based
filesystems properly emulated hard links, then I could simply point
rsnapshot to a compressed mountpoint to get compression like BackupPC...
Q2: Has anyone used the rsync option --checksum-seed with rsnapshot?
(I.e., to cache checksums on the server, the way BackupPC does?)
If those two things worked, then I could slap a generic PHP "File
Explorer" on my rsnapshot dir and have something that does everything
BackupPC does, except with backups that appear as standard files on
the filesystem (so I could do emergency restores using SSH).
--Derek
On 10/11/2009 08:28 AM, Mark Foster wrote:
William Kreuter wrote:
Thanks again to everyone who offered help with my
inquiries last month about off-the-shelf freeware
backup tools. I settled on rsnapshot and have it
running now.
What's the easiest way to tell which files have been
written afresh on each new backup? The target
volume for rsnapshot is, according to df, filling
slightly faster than I would have expected.
Billy
Bump the verbosity and the files backed up should appear in the
rsnapshot.log
Or you can run (something like) rsnapshot -t hourly
and find the backup set you are suspicious of , run the rsync command it
shows you (but possibly replace -q with -v and definitely -n) to get a
real-time explanation.
