Off the top of my head, it seems that the snapshots predate the directory itself. How does the snapdir look if you "cd .snapshot" and "ls -l" from a UNIX workstation?
If you see something like the following: 14:34:59:~/test$ ls -l .snapshot total 0 ---------- 1 root root 0 Sep 30 06:01 hourly.0 ---------- 1 root root 0 Sep 30 06:01 hourly.1 ---------- 1 root root 0 Sep 30 06:01 hourly.2 ---------- 1 root root 0 Sep 30 06:01 hourly.3 ---------- 1 root root 0 Sep 30 06:01 hourly.4 ---------- 1 root root 0 Sep 30 06:01 hourly.5 ---------- 1 root root 0 Sep 30 06:01 nightly.0 ---------- 1 root root 0 Sep 30 06:01 nightly.1 14:35:03:~/test$
Then that directory is newer than any of the snapshots. This makes the snapdir meaningless so we place zero permissions on the snaphot entries.
Again, that's all I got off the top of my head.
Steve Losen wrote:
We have a F630 running 4.2a and are having trouble accessing snapshots with CIFS.
Our snapshot schedule is hourlies every two hours, keep 8, and keep 7 nightlies, so we usually have 15 lying around.
We use both NFS and CIFS, with unix style file protection semantics.
From NT and W95, you can get into a share and use it just fine.
If you browse the ~snapshot dir, you see all the hourly.* and nightly.* entries, but if you click on any of them, you get an access denied error box.
If you "dir" the snapshot directory you see all the hourly.* and nightly.* entries, but if you dir any of those, they are empty.
The live directory has plenty of files in it and from NFS, you can access files in all the snapshots.
Again, the live filesystem works just as you would expect. It's only the snapshots that are giving us trouble.
By the way, I posted an earlier problem about the set-group bit and later realized that I was not a member of my test file's group, so of course I couldn't set the set-group bit - duh.
Steve Losen scl@virginia.edu phone: 804-924-0640
University of Virginia ITC Unix Support