Oracle releases Oracle Database 11g Release 2 (11.2.0.1.0) for Linux x86 and Linux x86-64 platforms.

If your are getting:
ORA-39125: Worker unexpected fatal error in KUPW$WORKER.UNLOAD_METADATA while calling DBMS_METADATA.FETCH_XML_CLOB [TAB
LE:"UNIUS"."SBA_E_SYSTEM_H"]
ORA-01801: date format is too long for internal buffer
during expdp or dbms_metadata exports, it’s because there are stored incorrect date values in the dictionary.

From my previous expriences with script utlrp.sql provided by Oracle, for mass-compilation of invalid objects, I recommend running this script only in STARTUP UPGRADE / MIGRATE mode. Otherwise you may end up with dictionary and or schema incosistencies especially if you use alot of schema objects (45 000+) .
$ORACLE_HOME/rdbms/admin/utlrp.sql

10.2.0.3 has a nasty bug, where one is not able to see what SQL is job actually executing, the following fix will resolve that partially. For parallel slaves the relevant v$session columns are still going to be empty.
5928612 SQL ADDRESS COLUMN IN V$SESSION VIEW FOR JOB IS ZERO
This might be resolved completely in 10.2.0.4, anyone?

If you wonder what is the password for Oracle Internet Directory(OID) user cn=orcladmin after OIM installation, well, it is exactly the same as ias_admin password specified during installation of OIM. If you really want to reset the password, use included utility resetiASpasswd.sh in $ORACLE_HOME\bin as follows:
First make sure your environment is setup properly for OIM [...]

Oracle patchset 10.2.0.4 is now available via metalink for Windows(32bit) and Linux x86(32bit) platforms.
Update 22.05.2008: Now available also for some 64 bit platforms like AIX5L 64.

If you’ve just upgraded from 9i to 10g, and you are using unsupported collection return types in pipelined functions you are no longer able to compile the code due to a more restrictive 10g compiler. Your pipelined functions compiled just fine on 9i. You are unable to compile the functions with the following error message [...]

Microsoft Windows (1.7 GB)
Linux x86 (1.7 GB)
Linux x86-64 (1.8 GB)

Too much log switching is no good for database, you should see numbers not bigger than 6 per hour. Increasing the size of redo log groups is one way to lower log switch frequency, please refer to Oracle documentation on how to lower log switch frequency.
You can use the following query to determine your log [...]

You can determine users idle time using the following query: