[rt-devel] New Installation Woes

[D. Hageman]

On Thu, 10 Jul 2003, Harald Wagener wrote:
> 
> On Donnerstag, 10. Juli 2003, at 8:27 Uhr, D. Hageman wrote:
> >
> > Below is a message that I posted to rt-users a couple of days ago with 
> > no
> > response.  I have since setup rt 3.0.3 on a RedHat 9 box and have
> > discovered the same issues.  Has anyone recently tried a clean install 
> > with
> > a similar setup?
> >
> > rt-users message:
> >
> > I have setup rt 3.0.3 on a clean installation of RedHat 7.3.  I am 
> > using
> > PostgreSQL 7.3.3, Perl 5.8.0, apache 1.3.27 and mod_perl 1.27  I have 
> > all
> > the latest modules updated via CPAN.
> 
> Remember to set the locale to a non-utf8-version, or some cpan modules 
> will compile incorrectly. This can be done in
> /etc/sysconfig/i18n
> for the complete system.  Also, the RedHat provided perl 5.8.0 is built 
> as multi threading, which is not recommended for production systems. My 
> recommendation is to rebuild perl from scratch. Some of the 
> RedHat-provided perl modules are known to be old and buggy/broken 
> versions - Encode.pm was last weeks favorite on the rt mailing lists.

I always set the LANG to be "C" before working with any perl.  I only use 
the base perl RPM from RedHat and use CPAN to install all the rest of the 
modules as CPAN rpms aren't exactly a convient way to maintain a system.

I also rebuild the RedHat RPM to remove the threading on my production 
systems before I use it.  

I have tried it with both the RedHat shipped items and the "recommended" 
method to no avail.

> > It seems that no combination of username and password (not even 'root' 
> > and
> > 'password' ) allows me to log into the RT system.
> 
> This is related to the MD5 module provided with RedHat's perl. See the 
> archives for details. My recommendation: Switch to a distribution that 
> is usable on productive server systems. RHAS might be usable, but I 
> won't shell out >1,000€ just to find out it does not cut my edge.

Since I don't use the MD5 shipped with RedHat's perl I can't believe this 
is the case.  I do appreciate your suggestions though.  

Let me give a little bit more evidence on that it is something else.  If 
you turn on PostgreSQL's statement logging I find stuff like this in the 
logs:

LOG:  query: SELECT  * FROM Users WHERE lower(Name) = 
'rt::currentuser=hash(0x85526f8)

Essentially, it appears for some reason the 'ref' code in 
DBIx::SearchBuilder is failing.  Vincent ... any ideas?

-- 
//========================================================\\
||  D. Hageman                    <dhageman at eecs.ku.edu>  ||
||  Information Specialist        1010C Learned Hall      ||
||  Phone: 785.864.3923                                   ||
\\========================================================//