GNU/Linux Desktop Survival Guide by Graham Williams
Desktop Survival Project Home Preface Advocacy History Distributions Installation Basic Survival Wajig Audio AbiWord APM Backup Bash Booting Cameras CD Chinese ChRoot Clock Commercial Conference Management Configuration CRON CVS Databases DIA Directories Disks Documentation DVD Email Eye of Gnome Evolution File Systems Firewalls Floppy Disks Fonts Ghemical GIMP Glade Gnome GnuCash Gnumeric GQView Graphics Groups HTML Editors ImageMagick Initialisations on Booting Java KDE Kernels Keyboard Konqueror KVM Switch LaTeX Log System Login Magellan Mathematics Memory Modems Mounting Devices MS/Windows Music MySQL Nautilus Networks NFS NT File System OpenOffice.org Oracle Packages Partitions Passwords PDF Photos PPP Presentations Printing Python R Remote Desktops Rsync Samba Scanning Science Security SH File System Sketch Spell SPlus Spruce SSH SUDO Subversion Swap Themes USB Video Virtualization Web Word X XML Sample Installations Troubleshooting Index

Installing Oracle 10g

Ensure X forwarding is enabled, and log in as user oracle:

xhost +
su oracle

Start the oracle installer:

cd /tmp/Disk1
./runInstaller

If everything is working you should see a stream test pass, and a window will pop up. If not, go back and make sure you didn't miss anything.

It is safe to ignore two rpm errors in the console.

With older versions of the libaio package installed (seems like the unstable version does not have this problem) two compile errors arise but can be ignored (in the GUI--just click continue). The two errors concern ``ihsodbc ctx_on'' and ``all_no_orcl.''

Before running /home/oracle/product/10.1.0/Db_1/root.sh as root, tell it to use the Debian idea of a nobody group (suggested by Martin.Schuster):

  - $CHGRP nobody $ORACLE_HOME/bin/extjob
  + $CHGRP nogroup $ORACLE_HOME/bin/extjob

Now run the root.sh script.

Pre 10.2 Oracle installed a service called cssd. The script waits 600 seconds (10 minutes) for the service to be started before going on. You need to do some set up while it waits for you! The message on the console will show: Expecting the CRS daemons to be up within 600 seconds. There are three things to do (as suggested by Eduardo Diaz Comellas).

First, the /etc/inittab file needs a line to set the active runlevels to 23 (the Debian defaults) instead of 35 (the Red Hat and Suse defaults). Change it from:

  h2:35:respawn:/etc/init.d/init.cssd run >/dev/null 2>&1 </dev/null

to

  h2:23:respawn:/etc/init.d/init.cssd run >/dev/null 2>&1 </dev/null

Next, the inittab runs /etc/rc.d/init.d/init.cssd. Exit this script and, at line 83, replace:

  SU="/bin/su -l"

with

  SU="/bin/su "

The Debian su does not support the -l option.

Finally, make init restart the service:

  # init q

The root.sh script should now continue and finish as expected.