Oracle® Database

4.1.1 Setting Kernel Parameters on Oracle Solaris 10

On Oracle Solaris 10, setting kernel parameters in /etc/system is deprecated. However, if you set shared memory using project settings, then you may encounter the following error:

ORA-27102: out of memory

Workaround:

Set memory, both by using allocations with project commands, and continue to provide memory settings in /etc/system. For example,

To set memory to 4 GB:

1. Open /etc/system with a text editor, and set the following values:

   set shmsys:shminfo_shmmax=4294967295
   set shmsys:shminfo_shmmni=100 
   set shmsys:shminfo_shmseg=10
   set semsys:seminfo_semmni=70
   set semsys:seminfo_semmns=200
   set shmsys:shminfo_shmmin=1
   

2. Set project memory settings in the kernel similar to the following:

   $ prctl -n project.max-shm-memory -r -v 4G -i project 3 
   $ projmod -sK "project.max-shm-memory=(privileged,4G,deny)" default
   

3. Create the project in /etc/project for the Oracle Database installation owner, oracle, and OSDBA group, dba, using projects -l so that the value persists after system restarts:

   % projects -l ... ... OASB projid : 3 comment: "" users : oracle groups :
   dba attribs: project.max-shm-memory=(privileged,4294967295,deny)
   

This issue is tracked with Oracle bug 8929233.

4.2.1 Oracle ASMFS Listed As a Storage Option When Installing Oracle RAC Database

If you install an Oracle RAC database after installing Oracle Grid Infrastructure for a cluster, then the Specify Database Storage Options screen indicates that "Automatic Storage Manager File System (ASMFS)" is a storage option for Oracle Database files.

Workaround:

Ignore this sentence. Either select a file system for storage, or select Automatic Storage Management.

This issue is tracked with Oracle bug 9032054.

4.2.2 Error When Installing Oracle Grid Infrastructure with the Grid Naming Service (GNS) and the OCR and Voting Disks Stored on Oracle ASM

When installing Oracle Grid Infrastructure for Oracle Database 11g Release 2 with GNS and the OCR and voting disks on Oracle ASM, root.sh may fail with error

PROT-1: Failed to initialize ocrconfig

The failing command is ocrconfig -upgrade user group, and TNS-12560 can be found in sqlnet log file.

Workaround:

Check the /etc/group entry for the group being used by OCR, temporarily reduce the number of users listed for this group, and restart the installation.

This issue is tracked with Oracle bug 8979500.

4.2.3 Upgrading Oracle RAC from Oracle9i to Oracle Database 11g Release 2 (11.2)

When you install Oracle Grid Infrastructure (Oracle Clusterware and Oracle ASM) on clusters created using Sun Clusterware and Oracle9i Oracle Real Application Clusters (Oracle RAC), then an issue on some systems may prevent the installer from detecting which nodes are cluster members of the existing cluster.

Workaround:

If the installer does not detect all nodes of the existing cluster, then you must select manually all the nodes where Oracle9i Oracle RAC is installed. If you select only a subset of the nodes that are part of your existing Oracle 9i Oracle RAC installation, then the existing Oracle 9i Oracle RAC installation will have problems.

This issue is tracked with Oracle bug 9040596.

4.2.4 Incorrect SSH Location on Installation Media

The SSH setup script, Disk1/sshsetup/sshUserSetup.sh, fails with the following errors as the installation media shows incorrect path:

./sshUserSetup.sh: /usr/local/bin/ssh: not found
./sshUserSetup.sh: /usr/local/bin/ssh-keygen: not found

Workaround:

Copy the executables to /usr/local/bin:

cp /usr/bin/ssh-keygen /usr/local/bin/ssh-keygen
cp /usr/bin/ssh        /usr/local/bin/ssh

This issue is tracked with Oracle bug 9044791.

4.2.5 Error Message Appears During Oracle ASM Upgrade to Oracle Grid Infrastructure

The rootupgrade.sh script displays the following error while upgrading Oracle ASM from release 10.1.0.5 to Oracle Grid Infrastructure release 11.2:

Failure 3 at CSS context initialization

Workaround:

Ignore this error message and proceed with the upgrade.

This issue is tracked with Oracle bug 8853996.

4.2.6 Deinstallation Tool Fails to Remove Oracle Database Gateway When Run from Oracle Home

Use the standalone version of the deinstallation tool to remove Oracle Database Gateway, instead of running the deinstallation tool from the Oracle home.

This issue is tracked with Oracle bug 9039543.

4.2.7 Trace Files Show Error Message After Oracle RAC Database Upgrade from Release 10.2.0.4 to Release 11.2

Oracle Database Upgrade Assistant reports the following error message in the trace files after upgrading Oracle RAC database from release 10.2.0.5 to release 11.2:

PRKR-1078 : Database new of version 10.2.0.0.0 cannot be administered using current version of srvctl.

Workaround:

Ignore this error message.

This issue is tracked with Oracle bug 8915673.

4.2.8 Oracle Database Vault Configuration Assistant (DVCA) Fails When Adding Languages

The command-line utility, DVCA, fails to add more languages to Oracle Database Vault giving the following error:

Exception in thread "main" java.lang.UnsatisfiedLinkError:
(path)/lib/libocijdbc11.so: ld.so.1: java: fatal:
(path)/lib/libocijdbc11.so: wrong ELF class: ELFCLASS64 

Solution:

Add the –d64 option when starting DVCA:

dvca -d64 -action addlanguages

This issue is tracked with Oracle bug 9151315.

4.3.1 Global Services Daemon Synchronization Issue

If Global Services Daemon is enabled, then it may fail to start on all nodes of a cluster when Oracle Grid Infrastructure is started or when the nodes are restarted. You can verify the daemon state by running the following command:

srvctl status nodeapps

Workaround:

If required, run the following command on any one node in Oracle Grid Infrastructure for a cluster to start the daemon on all the nodes:

srvctl start nodeapps

This issue is tracked with Oracle bug 9075123.

4.3.2 Help Files Translation Issue for Spanish and French

Oracle Universal Installer does not copy the translated help files to Oracle home directory when you select the Spanish or French language.

Workaround:

On the Select Product Languages page, select both the Latin American Spanish and Spanish to get the help files translated in Spanish; and select both the Canadian French and French for the translated help files in French.

This issue is tracked with Oracle bug 9147106.

4.3.3 Incorrect Reverse Ping Command from Oracle Management Service

Oracle Management Service provides incorrect reverse ping command with Oracle Enterprise Manager Database Control. The emoms.properties file shows:

emdrep.ping.pingCommand=/usr/sbin/ping <hostname>

Workaround:

Change the property in the emoms.properties file to:

emdrep.ping.pingCommand=/usr/sbin/ping <hostname> 30

This issue is tracked with Oracle bug 9211800.

4.3.4 Help Pages Fail to Load for Oracle Database Client (32-Bit)

Oracle Universal Installer does not find the clientinstaller_help.map file and fails to load the help pages for locales other than English.

Workaround:

Set LANG=C before running Oracle Universal Installer. This changes the installer locale to English and the help content is available.

This issue is tracked with Oracle bug 9474214.

5.1.1 Restarting the Installation Fails After Selecting Software Updates

Restarting the installation fails if the user chooses to restart after selecting the software updates option during an Oracle Grid Infrastructure installation on a computer with vendor clusterware.

Workaround:

Set the SKIP_ROOTPRE environment variable to TRUE before starting Oracle Universal Installer.

This issue is tracked with Oracle bug 10078957.

5.1.2 SSH Setup Issues

Oracle Universal Installer fails to setup SSH if there is any deviation from the standard SSH path.

Workaround:

1. Copy the /stage/properties/ssPaths_sol.properties file to your local computer.

2. Modify the relevant SSH path. For example, SSH_PATH=/bin/ssh.

3. Start Oracle Universal Installer and specify the location of the local properties file from step 1.

   runInstaller -J-Doracle.sysman.prov.PathsPropertiesLoc=local_path_location
   

This issue is tracked with Oracle bug 9837487.

5.2.1 Installing Predownloaded Software Updates

By default, software updates are download to the /tmp directory. This directory is deleted at the end of an installation. If you do not download and install the software updates during the same installation session, then the software updates are not available.

Workaround:

Specify a new download location for the software updates on the Download Software Updates screen.

This issue is tracked with Oracle bug 9975999.

5.2.2 Error Message Appears During Oracle ASM Upgrade to Oracle Grid Infrastructure

The rootupgrade.sh script displays the following error while upgrading Oracle ASM from release 10.1.0.5 to Oracle Grid Infrastructure release 11.2:

Failure 3 at CSS context initialization

Workaround:

Ignore this error message and proceed with the upgrade.

This issue is tracked with Oracle bug 8853996.

5.2.3 PRVF-5150 Error on Oracle Grid Infrastructure

When installing Oracle Grid Infrastructure, if the Oracle ASM device path is not specified in the format /dev/rdsk/cMtNdXsY or /dev/did/rdsk/dXsY or /dev/md/rawdg/rdsk/dX, then the following error displays during the Oracle ASM device validation prerequisite check:

PRVF-5150: Path /dev/rdsk/link/raw11 is not a valid path on all nodes

Workaround:

Specify the character device name rather than the symbolic link name.

This issue is tracked with Oracle bug 10052786.

5.2.4 ORA-27102 Out of Memory Error

When creating a database on a computer with large physical memory, Oracle Universal Installer or Database Configuration may display an out of memory error.

Workaround:

Select the Advanced Installation type during installation, or run Database Configuration Assistant after a software only installation and reduce the memory allocated for Oracle based on the shared memory settings for your operating system.

This issue is tracked with Oracle bug 9811726.

5.2.5 Global Services Daemon Synchronization Issue

If Global Services Daemon is enabled, then it may fail to start on all nodes of a cluster when Oracle Grid Infrastructure is started or when the nodes are restarted. You can verify the daemon state by running the following command:

srvctl status nodeapps

Workaround:

If required, run the following command on any one node in Oracle Grid Infrastructure for a cluster to start the daemon on all the nodes:

srvctl start nodeapps

This issue is tracked with Oracle bug 9075123.

5.2.6 Relinking Fails on 32-Bit Client Installations

Relinking fails on 32-bit client installations because the relink script tries to load 64-bit libraries.

Workaround:

After installation, manually remove the -d64 flag from JRE_MEMORY_OPTIONS in the $ORACLE_HOME/oui/oraparam.ini file.

This issue is tracked with Oracle bug 10027801.

5.2.7 Oracle Grid Infrastructure Fails to Start Automatically After a Node Restart

After you install Oracle Grid Infrastructure for a Cluster and Oracle RAC Database, if you restart one of the nodes, then Oracle Grid Infrastructure fails to start automatically.

Workaround:

Perform the following steps:

1. Add /usr/bin to the PATH environment variable.

2. Change "LC_CTYPE=en_US.UTF-8" to "LC_CTYPE=" in file, /etc/default/init.

   Or, comment the line:

   "#LC_CTYPE==en_US.UTF-8"
   

3. Restart the server.

This issue is tracked with Oracle bug 10124517.

5.2.8 Performance Statistics Issue with Oracle Enterprise Manager Database Control

The summary chart on the Oracle Enterprise Manager Cluster Performance page does not show any data.

Workaround:

Perform the following steps:

1. Stop Oracle Enterprise Manager Database Control.

2. Add the following to the properties file, $ORACLE_HOME/host_dbuniquename/sysman/config/emoms.properties:

   oracle.sysman.db.rac.useRemoteCollection=false
   

3. Restart the Oracle Enterprise Manager Database Control.

This issue is tracked with Oracle bug 10129539.

6.3.1 Passwordless SSH Issue During Oracle Clusterware Installation on Oracle Solaris 11

Passwordless SSH setup fails on Oracle Solaris 11 during the Oracle Clusterware installation.

Workaround:

Follow these steps before configuring SSH using Oracle Clusterware release 11.2.0.3 on Oracle Solaris on x86-64 (64-bit):

1. Edit /etc/ssh/sshd_config on the target hosts and add a cipher list which includes the CTR Ciphers and the longer AES key lengths before the CBC Ciphers. For example:

   Ciphers aes128-ctr,aes192-ctr,aes256-ctr,aes128-cbc,aes192-cbc,aes256-cbc,3des-cbc,blowfish-cbc
   

2. Restart the sshd daemon on the target host using the following command:

   # svcadm restart ssh 
   

This issue is tracked with Oracle bug 10039955.

Manually setup SSH connectivity between the nodes on Oracle Solaris on SPARC (64-bit).

This issue is tracked with Oracle bug 13035076.

6.3.2 Oracle Universal Installer Not Displayed When Installing Oracle Database on Oracle Solaris 11

When installing Oracle Database 11.2.0.3, Oracle Universal Installer may not appear on Oracle Solaris 11.

Workaround:

Set the AWT_TOOLKIT environment variable to XToolkit.

This issue is tracked with Oracle bug 12657154.

6.3.3 Operating System Locale Issue on Oracle Solaris 11

When installing on a Japanese locale (LANG=ja_JP.eucJP or ja_JP.pck), Oracle Universal Installer may show unnecessary characters on Solaris 11.

Workaround:

Either set the locale to Japanese using the UTF-8 encoding (LANG=ja_JP.UTF8) or use an English locale.

This issue is tracked with Oracle bugs 13707754 and 13853207.

6.3.4 Oracle Enterprise Manager Database Control Error During Upgrades

When upgrading Oracle RAC from release 10.2.0.5 to 11.2, the Oracle Enterprise Manager Database Control home page may show the following error:

java.lang.Exception: No Such Target - wait_bottlenecks

Workaround:

Run these emctl commands on the master agent. Go to the cluster database home page and click View All Properties link to identify the master agent host. Recompute dynamic properties for the Oracle RAC instance and Oracle RAC database on the master agent host using the following commands:

emctl reload agent dynamicproperties rac_instance_target_name:oracle_database 
emctl reload agent dynamicproperties rac_database_target_name:rac_database 

If the error persists, then increase the timeout for dynamic properties for the target by editing the emd.properties file located at $ORACLE_HOME/host_dbname/config, and providing a larger value for timeout_value. Increase the timeout value on all the nodes of the cluster. For example:

dynamicPropsComputeTimeout_oracle_database=timeout_value
dynamicPropsComputeTimeout_rac_database=timeout_value
emctl reload agent
emctl reload agent dynamicproperties rac_instance_target_name:oracle_database
emctl reload agent dynamicproperties rac_database_target_name:rac_database

This issue is tracked with Oracle bug 12907579.

6.3.5 CVU IPMP Group Consistency Check Failure

If you install Oracle Grid Infrastructure on a cluster that has Active-Passive IPMP configured, then the prerequisite check for Oracle Solaris IPMP group fail-over consistency may fail.

Workaround:

Ignore this failure and perform the following steps:

1. After installation, log in as root and use the following command syntax to add the standby NICs to the network resource:

   'srvctl modify nodeapps -S
    subnet/netmask[/if1[/if2|...]'
   

   where if1 and if2 are the IPMP NIC names.

2. Classify all network interfaces in the IPMP group as public by using the following command syntax:

   oifcfg setif -global if_name/subnet:public
   

   where if_name is the interface name and subnet is the subnet.

   For example:

   # srvctl modify nodeapps -S '192.0.2.0/255.255.255.0/eth3|eth5'
   # oifcfg setif -global eth3/192.0.2.0:public
   # oifcfg setif -global eth5/192.0.2.0:public
   

   Note:

   The default IP address of the standby NIC is 0.0.0.0. When you run the oifcfg command to classify the standby NIC as public, you must set the subnet to the same subnet as the active NIC.

This issue is tracked with Oracle bug 12939432.

6.3.6 Error Message Appears During Oracle ASM Upgrade to Oracle Grid Infrastructure

The rootupgrade.sh script displays the following error while upgrading Oracle ASM from release 10.1.0.5 to Oracle Grid Infrastructure release 11.2:

Failure 3 at CSS context initialization

Workaround:

Ignore this error message and proceed with the upgrade.

This issue is tracked with Oracle bug 8853996.

6.3.7 Global Services Daemon Synchronization Issue

If Global Services Daemon is enabled, then it may fail to start on all nodes of a cluster when Oracle Grid Infrastructure is started or when the nodes are restarted.

Workaround:

Run the following command on any one node in Oracle Grid Infrastructure for a cluster to start the daemon on all the nodes:

srvctl start nodeapps

This issue is tracked with Oracle bug 9075123.

6.3.8 Oracle Cluster Registry Integrity Check Fails After Oracle Grid Infrastructure Upgrade

When the Cluster Verification Utility performs the OCR check after upgrading Oracle Grid Infrastructure release 10.2.0.5 to release 11.2.0.3, the check may fail with the following error:

PRVF-4162 : Size of the OCR location "ocr_location" could not be
determined... 

Workaround:

Run the OCRCHECK utility as a privileged user to verify the contents of OCR.

This issue is tracked with Oracle bug 13011520.

6.4.1 OPatch Commands Give Warning Message

When you run OPatch commands, you may see the following warning message:

whereis: not found

Workaround:

Update the PATH environment variable by adding /usr/ucb.

This issue is tracked with Oracle bug 12793552.

6.4.2 Using Projects with Oracle Grid Infrastructure Systems

If you are installing Oracle Grid Infrastructure 11g Release 2 (11.2.0.3), and you want to use Oracle Solaris Projects to manage system resources, then you must ensure that all Oracle installation owners are using the same default project. Use the /usr/bin/projects -d command to check that the default Project for the Oracle Grid Infrastructure installation owner (grid) and the default Project for Oracle Database installation owners (oracle, oracle1, oracle2, and so on) are the same Project.

For example, if you have an Oracle Grid Infrastructure owner called grid, and you have two Oracle Database installation owners called oracle1 and oracle2, and the default Project for the grid installation owner is mygridproj, then enter commands similar to the following:

$ cd /usr/bin
$ projects -d grid
mygridproj
$ projects -d oracle1
mygridproj
$ projects -d grid
mygridproj

In the preceding example, the default project for the installation owners grid, oracle1 and oracle2 is mygridproj.

Refer to your Oracle Solaris documentation for more information about configuring resources using Projects.

This issue is tracked with Oracle bug 11776492.

7.3.1 Linking Issue on Oracle Solaris on x86-64 (64-Bit)

If you attempt to link Pro*COBOL applications on Oracle Solaris 10 update 10, then the application may not link, and the link editor may report the following error:

make: Fatal error: Command failed for target `build'

Workaround:

Upgrade to Oracle Solaris 10 Update 11.

Workaround 2:

Edit the object file before running the link editor as follows:

% elfdump -cN.eh_frame foo
 
     Section Header[4]:  sh_name: .eh_frame
         sh_addr:      0x400378            sh_flags:   [ SHF_ALLOC ]
         sh_size:      0x6b8               sh_type:    [ SHT_PROGBITS ]
         sh_offset:    0x378               sh_entsize: 0
         sh_link:      0                   sh_info:    0
         sh_addralign: 0x8
     % elfedit -e 'shdr:sh_type .eh_frame sht_amd64_unwind' foo
     % elfdump -cN.eh_frame foo
 .
     Section Header[4]:  sh_name: .eh_frame
         sh_addr:      0x400378            sh_flags:   [ SHF_ALLOC ]
         sh_size:      0x6b8               sh_type:    [ SHT_AMD64_UNWIND ]
         sh_offset:    0x378               sh_entsize: 0
         sh_link:      0                   sh_info:    0
         sh_addralign: 0x8

This issue is tracked with Oracle bug 15827517.

7.3.2 Oracle Enterprise Manager Database Control Fails to Start

Oracle Enterprise Manager Database Control may fail to start on Solaris 11 systems with high memory configuration, (12 GB or more) and may give errors during the following operations:

• Install Oracle Database software with Create Database option

• Create database using Oracle Database Configuration Assistant and select Oracle Database Control configuration

• Run Oracle Enterprise Manager Configuration Assistant to configure Oracle Enterprise Manager

• Run any emctl command

The following is the error:

Error occurred during initialization of VM
Could not reserve enough space for code cache

Workaround:

Set the environment variable, EM_SOL_JAVA to -d64 before performing any of the above operations.

This issue is tracked with Oracle bug 17246576.