Oracle® Database

5.1.1 Oracle Grid Infrastructure Installation Issue

When installing Oracle Grid Infrastructure on an IPv6 enabled host, root.sh may fail with the following errors:

CLSRSC-180: An error occurred while executing the command 
'/Grid/12.1.0/grid/bin/srvctl start nodeapps -n stuzg06-ipv6'
CLSRSC-287: FirstNode configuration failed 

Workaround:

Run the following command before running the root.sh file:

sudo autoconf6

This issue is tracked with Oracle bug 16523490.

5.1.2 Cluster Verification Utility (CVU) Prerequisite Check Issue

The CVU installer prerequisite check incorrectly flags the following operating system patches, though they are not required.

AIX 6.1

IV45070 - use alternate glink file when -bbigtoc is used
IV44766 - new version of glink using 2-instruction TOC accesses

AIX 7.1

IV45070 - use alternate glink file when -bbigtoc is used
IV30579 - new version of glink using 2-instruction TOC accesses

Workaround:

Ignore the error.

This issue is tracked with Oracle bug 17814858.

5.1.3 Automatic SSH Configuration Issue

You must install the Bash shell before performing automatic SSH configuration with Oracle Universal Installer.

This issue is tracked with Oracle bug 16044765.

5.2.1 Bad Address Issue on AIX 6.1 TL7 SP5

On AIX 6.1 TL7 SP5 or equivalent TL/SP combinations, the Cluster Health Monitor executable, osysmond.bin, fails with the following error when calling the internal API, perfstat_netinterface:

ERROR: errno=14
ERROR: errno means : Bad address 

Workaround:

Upgrade your system to AIX 6.1 TL7 SP6 or AIX 7.1 TL1 SP6.

This issue is tracked with Oracle bug 16697958.

5.2.2 OCLUMON Tool Issue

The Cluster Health Monitor command, oclumon dumpnodeview, may intermittently fail on some nodes with the following error:

CRS-09104: No data available for the given nodes

In addition, the daemon log file shows the following:

[    CRFM][1029]crfm_connect_to: Wait failed with gipcret: 16 for conaddr
[ GIPCNET][1286] gipcmodNetworkProcessAccept: slosdep :  Bad file number (9)

Workaround:

Rerun the oclumon dumpnodeview command on the failing node. If the command still fails, then restart Oracle Clusterware stack.

This issue is tracked with Oracle bug 16621227.

5.2.3 File Permissions Commands Do Not Work in a Snapshot

If you use the chmod or chown commands with the -R option in a snapshot, these commands do not work.

This issue is fixed in AIX Service Packs released with the last component, -1334. For example:

7100-02-03-1334

Patches are available for the chmod and chown commands. See the IBM web pages at

http://www-01.ibm.com/support/docview.wss?uid=isg1IV31203

http://www-01.ibm.com/support/docview.wss?uid=isg1IV31603

This issue is tracked with Oracle bug 16401138.

5.2.4 Role Permissions Issue with X Terminal

When starting a new X terminal session (or using the associated commands, such as uxterm and aixterm), the oracle_devmgt role necessary for the administration of Oracle Automatic Storage Management Cluster File System and Oracle Automatic Storage Management Dynamic Volume Manager is not properly set.

Workaround:

Start a new X terminal session, and run the following command:

$ swrole oracle_devmgt

This issue is tracked with Oracle bug 10040892.

5.2.5 Cluster Health Monitor (CHM) Command Issue

The CHM command does not display any of the I/O device metrics:

oclumon dumpnodeview -v 

Workaround:

None. Use the vendor operating system commands to display the I/O device metrics.

This issue is tracked with Oracle bug 16436459.

5.2.6 COBOL Precompiler Issue

When you compile Pro*COBOL applications using the SQL Plan Management options on 32-bit AIX-based systems, with high system load of the SQL commands that are running, the procob binary may stop, and fail to generate the COBOL file.

Workaround:

On 32-bit systems, relink the procob binary to use a separate memory model by using the option, -bmaxdata:0xD0000000/dsa during linking.

This issue is tracked with Oracle bug 16847656.

5.2.7 Issue Related to SQL*Module for Ada Demonstration Programs

If you provide trailing spaces in the passwords while logging into Modada applications, the following error occurs:

ORA-1017: Invalid username/password; login denied

This issue is tracked with Oracle bug 16996851.

5.2.8 Highly Available Virtual IP Fails to Start

Highly available virtual IP may fail to start when you run root.sh or rootupgrade.sh using sudo command or when you start Oracle Grid Infrastructure using the following command:

sudo crsctl start crs

Workaround:

Log in as root and then run the root.sh, rootupgrade.sh, or the following command:

sudo crsctl start crs

This issue is tracked with Oracle bug 16445624.

5.2.9 root.sh Fails to Start Oracle Cluster Synchronization Services (CSS)

The root.sh script fails to start ora.cssd if the library path environment variable, LIBPATH, in the root's .kshrc file does not contain the correct Grid_home/lib value.

Workaround:

Do not specify LIBPATH in the root's .kshrc file, or ensure that the correct Grid_home/lib value is included in the LIBPATH.

This issue is tracked with Oracle bug 16438264.

5.2.10 Cluster Health Monitor (CHM) Command Does Not Display Correct Metrics for Huge Pages

The CHM command does not display correct hugepagetotal, hugepagefree, and hugepagesize metrics:

oclumon dumpnodeview -v 

Workaround:

None. Use the vendor operating system commands to display the Huge Page metrics.

This issue is tracked with Oracle bug 16683232.

5.2.11 Cluster Health Monitor (CHM) Command Does Not Display Metrics for Swap Memory

The CHM command does not display swapfree and swaptotal metrics:

oclumon dumpnodeview

Or

oclumon dumpnodeview -v 

Workaround:

None. Use the vendor operating system commands to display the Swap metrics.

This issue is tracked with Oracle bug 16655032.

6.2.1 Automatic SSH Configuration Issue

You must install the Bash shell before performing automatic SSH configuration with Oracle Universal Installer.

This issue is tracked with Oracle bug 16044765.

6.2.2 Oracle Database Configuration Assistant Fails to Create a Database

Oracle Database Configuration Assistant fails to create a database from the working copy on Oracle ACFS snapshot if AIX OS level is less than 6100-09-03-1415 or 7100-03-03-1415.

Workaround:

Ensure that AIX OS level is 6100-09-03-1415 or 7100-03-03-1415 to create a database from the working copy Oracle ACFS.

This issue is tracked with Oracle bug 19584188.

6.2.3 Directory Permissions Issue

If you start Oracle Universal Installer for installing Oracle Database release 10.2.0.1, the permission for the logs directory in oraInventory changes causing problems for other OS users installing the software.

drwxrwx--- changes to drwxr-x---

Workaround:

Change the permissions back to drwxrwx--- for the logs directory after Oracle Database release 10.2.0.1 software installation.

This issue is tracked with Oracle bug 19391835.

6.2.4 Workload Partition Configuration Issue for Pseudo Terminals

While installing a single-instance database with Oracle Multitenant on a workload partitioned system, the following error displays when you start Oracle Database Configuration Assistant:

Database Configuration Assistance failed to detect options for this Oracle
Installation.
ORA-00443: background process "MMAN" did not start
 
Due to this failure some database Configuration Assistant features will not
be available. Do you want to continue?

Workaround:

Set the total pseudo terminals to unlimited using the smitty tool.

This issue is tracked with Oracle bug 19443754.

6.3.1 Role Permissions Issue with X Terminal

When starting a new X terminal session (or using the associated commands, such as uxterm and aixterm), the oracle_devmgt role necessary for the administration of Oracle Automatic Storage Management Cluster File System and Oracle Automatic Storage Management Dynamic Volume Manager is not properly set.

Workaround:

Start a new X terminal session, and run the following command:

$ swrole oracle_devmgt

This issue is tracked with Oracle bug 10040892.

6.3.2 File Permissions Commands Do Not Work in a Snapshot

If you use the chmod or chown commands with the -R option in a snapshot, these commands do not work.

This issue is fixed in AIX Service Packs released with the last component, -1334. For example:

7100-02-03-1334

Patches are available for the chmod and chown commands. See the IBM web pages at

http://www-01.ibm.com/support/docview.wss?uid=isg1IV31203

http://www-01.ibm.com/support/docview.wss?uid=isg1IV31603

This issue is tracked with Oracle bug 16401138.

6.3.3 Cluster Health Monitor (CHM) Command Issue

The CHM command does not display any of the I/O device metrics:

oclumon dumpnodeview -v 

Workaround:

None. Use the vendor operating system commands to display the I/O device metrics.

This issue is tracked with Oracle bug 16436459.

6.3.4 OCLUMON Tool Issue

The Cluster Health Monitor command, oclumon dumpnodeview, may intermittently fail on some nodes with the following error:

CRS-09104: No data available for the given nodes

In addition, the daemon log file shows the following:

[    CRFM][1029]crfm_connect_to: Wait failed with gipcret: 16 for conaddr
[ GIPCNET][1286] gipcmodNetworkProcessAccept: slosdep :  Bad file number (9)

Workaround:

Rerun the oclumon dumpnodeview command on the failing node. If the command still fails, then restart Oracle Clusterware stack.

This issue is tracked with Oracle bug 16621227.

6.3.5 Cluster Health Monitor (CHM) Command Does Not Display Metrics for Swap Memory

The CHM command does not display swapfree and swaptotal metrics:

oclumon dumpnodeview

Or

oclumon dumpnodeview -v 

Workaround:

None. Use the vendor operating system commands to display the Swap metrics.

This issue is tracked with Oracle bug 16655032.

6.3.6 Bad Address Issue on AIX 6.1 TL7 SP5

On AIX 6.1 TL7 SP5 or equivalent TL/SP combinations, the Cluster Health Monitor executable, osysmond.bin, fails with the following error when calling the internal API, perfstat_netinterface:

ERROR: errno=14
ERROR: errno means : Bad address 

Workaround:

Upgrade your system to AIX 6.1 TL7 SP6 or AIX 7.1 TL1 SP6.

This issue is tracked with Oracle bug 16697958.

6.3.7 COBOL Precompiler Issue

When you compile Pro*COBOL applications using the SQL Plan Management options on 32-bit AIX-based systems, with high system load of the SQL commands that are running, the procob binary may stop, and fail to generate the COBOL file.

Workaround:

On 32-bit systems, relink the procob binary to use a separate memory model by using the option, -bmaxdata:0xD0000000/dsa during linking .

This issue is tracked with Oracle bug 16847656.

6.3.8 Input/Output Error Displayed During Oracle Grid Infrastructure Shutdown If Any Oracle ACFS is Offline

Before shutting down Oracle Grid stack, if any Oracle ACFS is reported offline by the command, acfsutil info fs, or is reported stale by the command, srvctl status filesystem, then bring it online. If you are not able to bring the file system online, then restart the node. This ensures that no processes are active on an offline Oracle ACFS resulting in the input/output error.

When Cluster Ready Services Oracle Root Agent, orarootagent, attempts to clean up the error, it cannot determine which processes are still accessing the file system and so, cannot stop them. As a result, orarootagent cannot dismount the file system and clean up the active resources.

The fix is a revised AIX fuser utility from IBM that is part of all the Technical Levels of AIX 7.1. For AIX 6.1 users, Oracle recommends that you install the appropriate minimum Service Pack for your TL from this list:

TL4/SP11
TL5/SP7
TL6/SP6
TL7/any
TL8/any
TL9/any

This issue is tracked with Oracle bug 17357831.

6.3.9 INS-32131 Error with SSH

The specified root privilege password fails with INS-32131 error if the maximum number of failed authentication attempts in SSH during a specific time interval is low.

INS-32131: Specified user ('string') doesnt have permission to execute configuration script using Sudo option.

Workaround:

Increase the number of failed authentication or login attempts in SSH for a specific time interval.

This issue is tracked with Oracle bug 19596509.

6.3.10 Restriction for Oracle ACFS Mounts

Mounting more than 200 Oracle ACFSs may result in system failure and the following message may appear in the OKS log file:

Persistent log thread creation failed.

Workaround:

Do not attempt to mount more than 100 Oracle ACFSs.

This issue is tracked with Oracle bug 18422045.

6.3.11 Issue Related to SQL*Module for Ada Demonstration Programs

If you provide trailing spaces in the passwords while logging into Modada applications, the following error occurs:

ORA-1017: Invalid username/password; login denied

This issue is tracked with Oracle bug 16996851.