
===============================================================================
                    Oracle Server Management Readme File

                             Release 10.2.0.1.0
===============================================================================



1. Changes from Release 10.1.0.2.0
2. References
3. Known Problems/limitations

=================================
1. Changes from Release 10.1.0.2.0
=================================

A database can be configured to have a MANUAL CRS management policy, in which
case it will not be automatically started at node startup.  Also, a database
with a MANUAL policy will only be restarted by CRS once instead of five
times.  The default policy remains AUTOMATIC.  The policy is set using 
the -y flag to "srvctl add database" or "srvctl modify database".

The Oracle Clusterware log files are located in the following directories:
<CRS home>/log/<hostname>
<ORACLE_HOME>/log/<hostname>

Oracle CRS commands and APIs are now documented; see Chapter 14 of the
Oracle Clusterware and Oracle Real Application Clusters Administration 
and Deployment Guide

Oracle Clusterware only supports lower-case values for the AUTO_START 
resource attribute. The values as described in the Oracle Clusterware API 
chapter should be "always", "restore", and "never". 


=============
2. References
=============
For detailed information on setting up the cluster database for management by 
EM, refer to:

    Oracle Clusterware and Oracle Real Application Clusters Installation and 
      Configuration Guide 10g Release 2
    Oracle Clusterware and Oracle Real Application Clusters Administration 
      and Deployment Guide 10g Release 2

=============================
3. Known Problems/limitations
=============================

Do not change a hostname after CRS installation.   This includes
adding or deleting a domain qualification. Instead remove and re-add
the affected node using MetaLink note 276434.1
 
Bug 3377903: INST OFFLINE AFTER SHUTDOWN AND STARTUP FROM SQLPLUS 
DESCRIPTION: When shutting down an instance and immediately starting it 
again using SQL*Plus, sometimes the CRS resource for the instance does not 
get started, because it is still busy stopping. In this case, 
srvctl status instance -d <name> -i <sid>
will incorrectly report the instance not to be running.
WORKAROUND: Start the instance's CRS resource with 
srvctl start instance -d <name> -i <sid>

Bug 4417761: ONS ENVIRONMENT NEEDS TO BE SET TO CRS_HOME LOCATION
DESCRIPTION:  The TNS listener when started using lsnrctl  instead of by
srvctl, cannot communicate with ONS.  This causes the TNS listener to fail
to drop service handlers for instances on nodes that have failed, which it
usually does by receiving an ONS event notification.
WORKAROUND: When starting the TNS listener with lsnrctl, first set the
environment variable ORACLE_CONFIG_HOME to CRS home.  Alternatively, use
the "srvctl start listener" command to start the listener.

Bug 4241481: CRSD PASSES NULL AS TZ WHEN STARTING THE DATABASE 
DESCRIPTION: When the OS local timezone is different from that specified
by the Oracle user's TZ environment variable, any database instances
started by CRS will use the OS local timezone.  This is because CRS will
start the instance with an empty TZ, rather than the Oracle user's TZ.
WORKAROUND: There are three possible workarounds
1) modify /etc/localtime so that it provides the default timezone that you
want. This file controls the behavior in the case that TZ is undefined.
2) srvctl setenv database -d <name> -T TZ=<val> 
This will result in the given value of TZ always being used to start instances
of the given database.
3) Use sqlplus instead of srvctl to start the database instances

Bug 4420331: CRS NODEAPPS ARE NOT CREATED WHEN UPGRADING 10.1 CRS TO 10.2 CRS
DESCRIPTION: If one installs 10.1 CRS without installing a RAC database
home, and then upgrades to 10.2 CRS, the node applications do not get
created as part of the upgrade.  This is because the 10.1 node applications
run from a RAC database home, and are only created as part of 10.1 RAC
database install.
WORKAROUND: Run vipca from 10.2 CRS home to add node applications for each node.

Bug 4409534: WARNING MESSAGE OF VIP IS ALREADY USED DURING CRS ADD NODE
DESCRIPTION: When adding a node and running root.sh on the new node, you will
get an error message "IP address "<VIP address>" has already been used. Enter
an unused IP address". Ignore this message as VIP for the new node already
got configured when rootaddnode script was run.
WORKAROUND: Ignore this message.

Bug 4407977: WARNING ERROR WHILE COPY DIR TO REMOTE NODE DURING CRS UPGRADE
DESCRIPTION: When upgrading 10.1.x to 10.2 CRS OUI may display a warning with 
the following error message "WARNING: Error while copying directory <directory>
with exclude file list <filename> to nodes <node>. [PRKC-1002 : All the
submitted commands did not execute successfully]" with list of all the files
which could not be copied.
WORKAROUND: If you see 'core' files is listed in the list of files not copied
then remove the listed core files and retry the upgrade

Bug 4411708: PROT-22 STORAGE TOO SMALL WHEN REPLACING A FAILED OCR DISK 
DESCRIPTION: During online OCR device replacement, ocrconfig might report 
the error "PROT-22: Storage too small".
The ocrconfig might report this error because OCR makes a conservative 
size estimate.  In some cases, the OCR might detect that the new replacement
location is not large enough to mirror the OCR.
WORKAROUND: To successfully perform an online OCR replacement, the new 
replacement location must be slightly larger than the size of the 
original location when the OCR was first created. If the new OCR location 
is a file, then the OCR will ensure that there is enough free space on the 
file system.
For example :
If the original device size is 120M, then the replacement device must
be at least 128M in size, or the replacement file should have at
least 128M of free space on the file system.
If the original device size is 160M, then the new replacement device must
be at least 256M in size, or the replacement file should have at least 
256M of free space on the file system.
In all cases, the required size of the new replacement device is no more 
than 128M larger than the size of original device. 

Bug 4431939: OCRCONFIG CMD DOES NOT RETURN ZERO EVEN IF IT SUCCEEDS 
DESCRIPTION: During online OCR device replacement or addition, ocrconfig fails
at the last step. There is no failure message displayed to the user.  The
ocrconfig tool reports this error because it fails to correctly set skgfr 
block0 on the new device/file. The failure message is logged in 
$ORA_CRS_HOME/log/<hostname>/client/ocrconfig_<pid>.log
WORKAROUND: Run $ORA_CRS_HOME/bin/clsfmt ocr <new ocr location>
For example :
./ocrconfig -replace ocrmirror /oradbshare/mirror.ocr

The log file for this invocation will have :
csetskgfrblock0: clsfmt returned with error [4]. 
<timestamp>: [ OCRCONF][<tid>]Failure in setting block0 [-1] 
<timestamp>: [ OCRCONF][<tid>]OCR block 0 is not set ! 
<timestamp>: [ OCRCONF][<tid>]Exiting [status=failed]... 

To fix this, issue the following command :
./clsfmt ocr /oradbshare/mirror.ocr

Cluster Installation
--------------------

1. Pre-Installation Setup
-------------------------

Before invoking the Oracle Universal Installer to start Oracle10g Enterprise
Edition installation for CRS or RAC, perform the following steps.

a. Set up the oracle user equivalence on all nodes of the cluster, including
   the local node.

b. If the file /var/opt/oracle/srvConfig.loc exists, ensure that the user
   performing the installation has permissions to read it.  If not, change
   the permissions with
   % chmod 644 /var/opt/oracle/srvConfig.loc
   The group of this file should be the OSDBA group.

c. Stop any 9.2 or earlier GSDs, if present

