The .snapshot directory in the root of the volume (or qtree, in most cases, IIRC) are where you need to start if you want to see the entire filesystem just as it was at snapshot time.
The .snapshot directories in subdirectories are there for the convenience of users to recover from 'Oh, #^%$' moments after deleting or modifying a file. Some changes in directory structure are not recoverable at this level, as you've seen. I really like having the .snapshot directories all over the filesystem, it's easier for users to recover their files. Systems folk can go to the .snapshot directory in the root of the volume to get the consistent entire-filesystem snapshot. - Richard Edward Ned Harvey wrote: > Did you do a "mv" somewhere in there? Or just forget to paste it? > > I'm still biased to think it's probably fine. Although Jonathan had one > test which produced a bad result, the assumption was that it would have been > good. And given that Adam had a good result, it's probably just something > weird and specific to something that Jonathan unfortunately has right now > ... > > Still, Tom, there's no harm in actually doing the test, if you'd care to > repeat it or paste it or whatever. Cuz it's not shown below. > > > >> -----Original Message----- >> From: Tom Limoncelli [mailto:[email protected]] >> Sent: Friday, April 23, 2010 5:08 PM >> To: Adam Levin >> Cc: Edward Ned Harvey; [email protected] >> Subject: Re: [lopsa-tech] Anybody with a netapp? >> >> t...@goodfor:~$ mkdir tmp/snaptest >> t...@goodfor:~$ cd tmp/snaptest/ >> t...@goodfor:~/tmp/snaptest$ mkdir http_root >> t...@goodfor:~/tmp/snaptest$ mkdir http_root/current >> t...@goodfor:~/tmp/snaptest$ mkdir http_root/current/bin >> t...@goodfor:~/tmp/snaptest$ echo "stable" > >> http_root/current/bin/myapp >> t...@goodfor:~/tmp/snaptest$ mkdir http_root/experimental >> t...@goodfor:~/tmp/snaptest$ mkdir http_root/experimental/bin >> t...@goodfor:~/tmp/snaptest$ echo "unstable" > >> http_root/experimental/bin/myapp >> >> t...@goodfor:~/tmp/snaptest$ find .snapshot/hourly.0 >> .snapshot/hourly.0 >> .snapshot/hourly.0/http_root >> .snapshot/hourly.0/http_root/current >> .snapshot/hourly.0/http_root/current/bin >> .snapshot/hourly.0/http_root/current/bin/myapp >> .snapshot/hourly.0/http_root/experimental >> .snapshot/hourly.0/http_root/experimental/bin >> .snapshot/hourly.0/http_root/experimental/bin/myapp >> >> (OH NO! My new version is unstable! I want to figure it out and go back >> to >> my hourly backup!) >> >> You just need to start one directory up from where you did the "mv": >> >> t...@goodfor:~/tmp/snaptest$ cat >> http_root/.snapshot/hourly.0/experimental/bin/myapp >> unstable >> t...@goodfor:~/tmp/snaptest$ cat >> http_root/.snapshot/hourly.0/current/bin/myapp >> stable >> >> netapp snapshots give you your data in the same place that zfs gives >> it to you PLUS additional paths. You said that zfs puts it in the >> right place. That means that netapp is putting it in the right place >> too... with additional places for your data restoration convenience. >> >> Tom >> > > _______________________________________________ > Tech mailing list > [email protected] > http://lopsa.org/cgi-bin/mailman/listinfo/tech > This list provided by the League of Professional System Administrators > http://lopsa.org/ > _______________________________________________ Tech mailing list [email protected] http://lopsa.org/cgi-bin/mailman/listinfo/tech This list provided by the League of Professional System Administrators http://lopsa.org/
