rm -r recursion failure

[John W. MacDonald]

This report is very similar to earlier ones involving recursion failure using
'rm -r' 
on directories on nfs mounted volumes. (see
http://lists.gnu.org/archive/html/bug-core
utils/2004-03/msg00016.html). In this case:

client: RedHat Enterprise WS 3.x, (kernel 2.4.21-15.0.4.ELsmp)

server: Apple Xserve OS X server v.10.3.8; filesystems: HFS+ (journaled) [tested
on ra
id volumes and internal drives]

rm version: rm (coreutils) 5.2.1

In our case, we are setting up this new XServe raid and discovered this
irritation whe
re removing study directories filled with medical images. The target directories
to be
 removed have one or two levels of subdirectories but a total of around 5000
files+. R
unning rm -r <targetdir> returns messages that none of the subdirectories have
been re
moved because they are not empty.

Inspecting the directories at this point shows that _some_ files have been
removed fro
m most or all of the directories, though generally a small percentage. Repeating
the r
m -r command five to ten times will eventually remove the target.

Doing a more simple test:

mkdir -p firstdir/seconddir/thirddir/fourthdir

Placing one or two files in any level and the entire target will be removed with
rm -r
.

But placing 48 typical files in fourthdir, running remove stops after removing
no file
s and 11 files in two separate tests.

Placing 48 files in firstdir, and rm -r removed 33 files in one test, as well as
all t
he empty subdirectories.

I have read somewhere that Apple's HFS+ filesystem is problematic for NFS
sharing with
 certain platforms. Apple tech support provided me with a document claiming a
problem 
with the remove algorithm.

In our case, the problem is 100% reproducible, and also occurs from SGI clients
runnin
g IRIX 6.5.16.

Any help or advice would be most welcome.

Appreciatively

John MacDonald
Dartmouth Medical School

address@hidden