Re: ANN: Backup Bouncer -- a metadata test suite
Re: ANN: Backup Bouncer -- a metadata test suite
- Subject: Re: ANN: Backup Bouncer -- a metadata test suite
- From: Nathaniel Gray <email@hidden>
- Date: Wed, 2 May 2007 15:52:34 -0700
Axel Luttgens wrote:
After having extracted the archive on the Desktop, I just ran those
commands:
cd ~/Desktop/backup-bouncer-0.1.1
make
./bbouncer create-vol Src
./bbouncer create /Volumes/Src
./bbouncer create-vol Dst
./bbouncer copy -d -c 10-rsync-apple /Volumes/Src /Volumes/Dst
without having changed anything in your code (of course, I first
checked it wasn't some kind of bad April joke ;-) ).
Thes are the results I get here:
------------------ rsync-apple ------------------
Verifying: basic-permissions ... ok
Verifying: timestamps ...
Sub-test: modification time ... ok
ok
Verifying: symlinks ... ok
Verifying: symlink-ownership ... ok
Verifying: hardlinks ... ok
Verifying: resource-forks ... ok
Verifying: finder-flags ... ok
Verifying: finder-locks ... FAIL
Verifying: creation-date ... FAIL
Verifying: bsd-flags ... FAIL
Verifying: extended-attrs ...
Sub-test: on files ... ok
Sub-test: on directories ... ok
Sub-test: on symlinks ... FAIL
FAIL
Verifying: access-control-lists ...
Sub-test: on files ... ok
Sub-test: on dirs ... ok
ok
Verifying: fifo ... ok
Verifying: devices ... ok
Verifying: combo-tests ...
Sub-test: xattrs + rsrc forks ... ok
Sub-test: lots of metadata ... ok
ok
on an Intel box as well as a PPC one.
Ah! Interesting! I was hoping to find someone like you. This means
I have to tell a story.
I didn't develop BB entirely for fun. The drive in my PBG4 started
developing bad blocks and I needed to replace it. ASR fails on I/O
errors, so I started thinking about file-by-file archival tools.
(However, in the end I used dd_rescue -- I recommend it highly.) The
plasticsfuture blog post from a year ago[1] left me full of doubt
about most tools, so I decided the only way to be sure of what would
happen would be to do my own testing.
As I developed BB I found that the results were not so bad. My
results looked like what you see above, and I got the same results
for tar. So I decided to go ahead and use rsync as my backup tool,
thinking it had been fixed. I booted from the Tiger install DVD,
mounted both my drives, and started my copy, being sure to use the
path to the up-to-date rsync on my system drive. The result? My
copy was a disaster. No finder flags, no resource forks. I tried
again with tar and got the same result. I tried using
DYLD_LIBRARY_PATH to make sure dynamic libraries were loaded from the
system drive as well, but still no luck. And all the while, my unit
tests were telling me it should all just work.
Until they didn't. On the evening of April 24 (according to the svn
logs) I was doing my best to trigger the failures I'd seen when
actually backing up. I tried disabling ACLs on my BB source volume
(to mimic my root volume), but nothing happened so I re-enabled
them. I did some minor tweaking of my code, including adding a test
for ACLs on directories and symlinks, but nothing that should have
had a big impact. Suddenly, the tests started failing for me, and I
couldn't find any way to get them to succeed again. I tried
reverting the code and starting with fresh disk images, but no matter
what I did, I got the failures. I would have thought this was a
*bad* thing, but my results now matched the plasticsfuture results,
and fresh checkouts on two colleagues' intel macs gave the same results.
So I have no idea why rsync/tar work on some systems and not on
others. One potential hint has to do with ACLs. BB tests copying
from a volume with ACLs enabled to another volume where they're
enabled. However, as I mentioned, I realized that they're *not*
enabled on my root volume, so I tried disabling them on my BB source
volume. Nothing happened -- the tests still passed. However, it was
soon thereafter that they failed (within an hour or so).
Did you get your own results on a clean 10.4.9 install?
No, not at all. I'm using my hard-working laptop. I don't have
access to a clean 10.4.9 install. However, my results seem to be
replicable on a variety of 10.4.9 systems.
Cheers,
-n8
P.S. Please CC me with replies -- I don't get mail from the list.
[1] - http://blog.plasticsfuture.org/2006/03/05/the-state-of-backup-
and-cloning-tools-under-mac-os-x/
--
>>>-- Nathaniel Gray -- Caltech Computer Science ------>
>>>-- Mojave Project -- http://mojave.cs.caltech.edu -->
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden