RE: Error: du: failed for admin: Cycle detected - toasters

29 Jul 2003

      Here's a quick Perl script I wrote to avoid .snaphost.
Hack it to your heart's content.  I call it mydu.  Simple
but effective.
#!/bin/perl
#
# Usage mydu [dir]
#
opendir (DIR, $ARGV[0]) || die "Can't open $ARGV[0] $!";
@dirs = readdir(DIR);
closedir (DIR);
foreach $d (@dirs)
{
  next if ($d eq "." || $d eq ".." || $d eq ".snapshot");
  system ("du -s $ARGV[0]/$d");
}
-- Adam Fox
NetApp Professional Services, NC
adamfox@netapp.com
-----Original Message-----
From: Chris Thompson [mailto:cet1@cus.cam.ac.uk]
Sent: Tuesday, July 29, 2003 2:18 PM
To: Art Hebert
Cc: toasters@mathworks.com
Subject: Re: Error: du: failed for admin: Cycle detected
art@arzoon.com (Art Herbert) writes:
...
I am getting this error when I try to do a  "du -sk *" some netapp
directories mounted on my solaris 2.9
You mean Solaris 9 (SunOS 5.9), presumably.
...
                                  oracle database server. If I am in the

directory that contains the files it works fine. If I try to do it one level
up I get the Cycle detected error:
sjcoracle01[root]202# cd /export/home/oracle/oradata/server01/admin
sjcoracle01[root]203# du -sk *
35692   RMAN
44      network
530080  prod01
sjcoracle01[root]204# cd ..
sjcoracle01[root]205# du -sk *
du: failed for admin: Cycle detected
du: failed for arch1: Cycle detected
du: failed for arch2: Cycle detected
du: failed for data1: Cycle detected
du: failed for data2: Cycle detected
du: failed for rbs1: Cycle detected
du: failed for redo1: Cycle detected
du: failed for redo2: Cycle detected
du: failed for temp1: Cycle detected
Mounted as follows:
/export/home/oracle/admin on sjcf760-01:/vol/vol0/oracle/admin
remote/read/write/setuid/hard/intr/vers=3/proto=tcp/forcedirectio/xattr/dev=
4580001
Anybody ever see this error?
Not until I did some experiments ...  :)
The "Cycle detected" message is apparently (from "strings" output) new
in Solaris 9 "du". It seems that it happens whenever one applies "du" to
a directory tree in which there are "visible" .snapshot directory entries
(i.e. one or more NFS mount points).
Presumably it's getting upset by the duplicate inode numbers appearing
in snapshots. Lots of Unix utilities (especially GNUish ones) have had
similar problems in the past.
[Of course, it's not unreasonable to take the line that their confusion 
is quite justified, and that NetApp have a design fault here.]
Chris Thompson
Email: cet1@cam.ac.uk