Hi Bernard,
 
I don't know how $PERL5LIB affects the installation. The "MySQL problem" in fact relates to $PERL5LIB. Simply modifying the paths in /etc/profile.d/perl5lib-profile-oscar.sh and run ./install_cluster again is not enough.
 
I rebuild a perl5lib-profile-oscar-1.0-2.noarch.rpm using perl5lib-profile-oscar-1.0-2.src.rpm with the perl5lib-profile-oscar.spec modified as follow:
 
For sh:
if ! echo \$PERL5LIB | /bin/grep -q "vendor_perl/5.8.0" ; then
   PERL5LIB=/usr/lib64/perl5/vendor_perl/5.8.0/x86_64-linux-thread-multi:\$PERL5LIB
   PERL5LIB=/usr/lib64/perl5/vendor_perl/5.8.0:\$PERL5LIB
   PERL5LIB=/usr/lib/perl5/vendor_perl/5.8.0:\$PERL5LIB
   export PERL5LIB
fi

For csh:
if ( "\$?PERL5LIB" ) then
    if ( "\${PERL5LIB}" !~ *vendor_perl/5.8.0* ) then
        setenv PERL5LIB "/usr/lib64/perl5/vendor_perl/5.8.0/x86_64-linux-thread-multi:\${PERL5LIB}"
        setenv PERL5LIB "/usr/lib64/perl5/vendor_perl/5.8.0:\${PERL5LIB}"
        setenv PERL5LIB "/usr/lib/perl5/vendor_perl/5.8.0:\${PERL5LIB}"
    endif
else
    setenv PERL5LIB ""
    setenv PERL5LIB "/usr/lib64/perl5/vendor_perl/5.8.0/x86_64-linux-thread-multi:\${PERL5LIB}"
    setenv PERL5LIB "/usr/lib64/perl5/vendor_perl/5.8.0:\${PERL5LIB}"
    setenv PERL5LIB "/usr/lib/perl5/vendor_perl/5.8.0:\${PERL5LIB}"
endif
Then the paths are correctly set during ./install_cluster eth1 , and the MySQL problem is gone.
 
I successfully run the oscar server installation. 
 
Now I have problem to build client image. I use the fedora-4-i386.rpmlist plus the last 14 lines of fedora-3-x86_64.rpmlist to make a fedora-4-x86_64.rpmlist . When I build client image, it stops at: (also see attached oscarinstall.log)
ERROR: The following RPMs cannot be installed:
 lam-oscar-7.0.6-4.fc4.i386.rpm: requires libc.so.6(GLIBC_2.1.3) (UNKNOWN)

I find libc.so.6 in /lib, and it is linked to libc-2.3.5.so .
What sh ould I do now? Should I build a "lam-oscar-7.0.6-4.fc4.x86_64.rpm".  I also find
lam-7.1.1-3.x86_64.rpm in the FC4 CD. Can I directly use that?
 
Thank you
Andy

Bernard Li <[EMAIL PROTECTED]> wrote:
Looks like the database is correctly created at least.  What happens when you re-run /opt/oscar/install_cluster eth0 ?
 
If that doesn't work, I suggest you run /opt/oscar/scripts/start_over and start over.  You probably want to make a backup of /tftpboot/rpm as the start_over script might delete some RPMs from that directory, so if you make a backup of it, you don't have to find the RPMs from the CDs and copy them over again.
 
Cheers,
 
Bernard

Attachment: oscarinstall.log
Description: 3086633976-oscarinstall.log

Reply via email to