Argghhhh! (pulling hair out of head)
What in blazes is going on? This is getting really strange. From the
shell prompt I can enter:
ls -la /WWW/WWW/httpd_1.3/htdocs/network/gif
and receive a listing of files in the "gif" subdirectory.
However, if from the CI prompt (logged on as the same user) I enter a
LISTFILE command I get a security violation:
listfile /WWW/WWW/httpd_1.3/htdocs/network/gif/@,2
Open of directory "/WWW/WWW/httpd_1.3/htdocs/network/gif" failed.
Security violation. (CIWARN 1080)
Okay, why can I view the directory contents from the shell, but not from
the CI? Oh, one thing to note. This is a symbolic link (which I'm about
ready to remove and replace with a "real" subdirectory - maybe that will
work). The really odd thing is that this all used to work (before the
RELOAD) just fine.
Also, both the NCSA and Apache web servers can serve files from the "gif"
directory, but indexing (displaying the list of files in the "gif"
directory" fails with a 403 error - Forbidden. So, why can MGR.WWW user
(interactively) do an "ls -la" on the directory and get a listing, but the
NCSA web server (which logs on as MGR.WWW) NOT be able to get a listing,
and why can SERVER.APACHE interactively from the shell do an "la -la" on
the directory, but have it fail when run from the server (also logged on as
SERVER.APACHE)?
The HP 9000 guru suggested that there may be a "traverse directories"
problem, but I wasn't aware that TD was a separate attribute that could be
set. I'm under the impression that if you have one of: read, write,
execute then you have TD. Is this correct? If not, how is TD specified?
And another oddity has appeared. All the cgi scripts that used to work
with both the NCSA and Apache web servers will now fail on the Apache web
server with an error 500. Upon looking in the Apache error_log I find that
the error is a "malformed header". Funny, it looks fine to me!
I'm now becoming very upset about RELOAD. I *used* to believe that as long
as you had a good SLT and full backup, you could recover your system to the
state it was in at the time of the full backup. Apparently this is not the
case. Just what is wrong, I don't know, but the only POSIX application
that seems to be working properly is Samba/iX.
It looks like I've got until Wednesday to get the web servers back up on
the HP 3000. If they are not back in full operation by 2 PM Wednesday, we
will move the web server over to an HP 9000 and shut down the servers on
the HP 3000. I understand why, after all, when they duplicated the
directory structures and moved the files over to the HP 9000 last Friday,
everything came right up on the HP 9000 with the exception of accessing
some reports that are produced on the HP 3000.
--------------------------------------------------------------
John Korb email: [log in to unmask]
Innovative Software Solutions, Inc.
The thoughts, comments, and opinions expressed herein are mine
and do not reflect those of my employer(s), or anyone else.
|
