3 Readme Information for Oracle Database 12c Release 1 (12.1.0.1)

3.22.1.1 Fine Grained Auditing (FGA)

If Fine Grained Auditing (FGA) is enabled on a table in the query, then Query Rewrite will not occur for this query.

3.22.1.2 Use of PARTITION Clause

Query rewrite does not occur for queries that use the PARTITION clause in the FROM clause to access table partitions. In order for query rewrite to rewrite such queries, the PARTITION clauses must first be converted into equivalent selection predicates and added to the WHERE clause.

3.24.1.1 Bug 16993778

Any time that you unlock an account (ANONYMOUS, in particular) in the ROOT database, you must then explicitly lock that account in any pluggable database (PDB) to which you do not want it to have access.

Workaround: None.

3.24.1.2 Bug 16682595

When a PDB is unplugged from a CDB, the values of the initialization parameters that were specified for the PDB with SCOPE=BOTH or SCOPE=SPFILE are added to the PDB's XML metadata file. However, these values are not restored for the PDB when it is plugged in to a CDB.

Workaround: None.

3.24.1.3 Bug 16660797

The failover DDL operation ALTER DATABASE FAILOVER TO <target_standby_db_unique_name> may fail to terminate all sessions connected to the standby database in a reasonable period of time. In such cases, the failover fails with the ORA-3113 error.

Workaround: Shut down all nodes at the standby database; then mount one node and rerun the failover DDL. If the failover operation continues to fail, then refer to steps 8 and 9 in Section 9.2.2 "Performing a Failover to Physical Standby" of Oracle Data Guard Concepts and Administration.

3.24.1.4 Bug 16654392

If you upgrade your database to Oracle Database 12c Release 1 (12.1) and then plug it into a CDB, the automatic SQL Tuning Advisor generates the following error:

ORA-65040: operation not allowed from within a pluggable database

This occurs because, in a 12.1 CDB, the automatic SQL Tuning Advisor is supposed to run at CDB$ROOT and not within a PDB. The error occurs because the PDB is automatically executing an old program which comes from the upgraded database. The problem does not exist for a brand new PDB.

Workaround: Connect to the PDB where you see the ORA-65040 error for the automatic SQL Tuning Advisor, use the DBMS_SCHEDULER.DROP_PROGRAM('AUTO_SQL_TUNING_PROG') procedure to delete the existing old AUTO_SQL_TUNING_PROG program, and then run the execsqlt.sql script to re-create it. You can find the script in the ORACLE_HOME/admin directory.

Automatic SQL Tuning Advisor can be configured at the CDB level only.

You can configure the following automated maintenance tasks at the CDB or PDB level:

• Optimizer Statistics gathering

• Segment Advisor

3.24.1.5 Bug 16624893

Shared servers cannot be enabled or disabled at the pluggable database (PDB) level. Therefore, either all PDBs or none of the PDBs must use shared servers unless one of the following workarounds is used.

Workaround: Any of the following workarounds can be used if only some PDBs in a multitenant container database (CDB) should use shared servers:

• Modify the connect alias for the connections that should be dedicated to include (SERVER=DEDICATED) in the CONNECT_DATA section.

• If the connections that should be dedicated share a sqlnet.ora initialization parameter file, set USE_DEDICATED_SERVER=on.

• In the dispatcher configuration, use the SERVICE= field to specify the list of PDB services that should use shared servers.

3.24.1.6 Bug 16618942

Pluggable database (PDB) cloning using a snapshot copy fails on Oracle ASM Cluster File System (ACFS) if schema files are present.

Workaround: None.

3.24.1.7 Bug 16274018

In a Standard Edition multitenant container database, unplugging PDB using the RMAN option in Database Configuration Assistant (DBCA) fails with the following error:

Error while while creating backup piece

Workaround: Use the PDB archive (TAR format) option or manually unplug the PDB using SQL*Plus.

3.24.1.8 Bug 16244949

In a CDB where the CURSOR_SHARING=FORCE parameter is set and a PDB is plugged in with a character set that is incompatible to the CDB, an attempt to open the PDB fails with the ORA-41400 error and the PDB is implicitly closed again. The incompatible character set violation is recorded in the PDB_PLUG_IN_VIOLATIONS view.

Workaround: None.

3.24.1.9 Bug 16074818

The default memory allocated to Oracle from DBCA is 40% of the total RAM. This may not be sufficient when creating a multitenant container database (CDB) with a higher number of pluggable databases (PDBs). With the default memory parameters, you may run into the ORA-4031 error while creating PDBs.

Workaround: Change the system global area (SGA) target to a larger value when using DBCA to create a CDB. Oracle recommends an increase to the default SGA of 24 megabytes multiplied by the number of PDBs. Higher values may be necessary depending on the workload.

3.24.1.10 Bug 16074307

After flashback CDB across PDB point-in-time recovery, temporary files may get deleted.

Workaround: Re-create the temporary files.

3.24.1.11 Bug 16049373

When creating a CDB with 252 PDBs using the default parameters, the ORA-00020 error is returned.

Workaround: Change the process number to a larger value when using DBCA to create a CDB.

3.24.1.12 Bug 15974047

Per-PDB session counts or limits are not sent to the listener. Thus, it is possible for a listener to direct a connection to an instance where the PDB is at the limit while the PDB on a different instance is not at the limit.

Note that this only matters if the PDB is at its session limit. If it is not, there is no impact from this bug.

Workaround: Do not use per-PDB session limits.

3.24.1.13 Bug 15965021

In Oracle Database 12c, you should choose between standalone Oracle Label Security (OLS) and OLS-OID (Oracle Label Security-Oracle Internet Directory) configuration for an Oracle Database 12c CDB while creating a CDB using DBCA. The OLS-OID configuration can be chosen by selecting the OLS-OID option while configuring Label Security in the Custom Install option. Once chosen, you cannot change the option for the lifetime of the CDB.

Also, you should not plug a standalone OLS-enabled PDB into a CDB which has OLS-OID enabled or vice versa. If done, warnings will be logged in the PDB_PLUG_IN_VIOLATIONS view. The incompatibility could be resolved by running rdbms/admin/catolsrecomp.sql.

Workaround: To remove the warnings in the PDB_PLUG_IN_VIOLATIONS view, refer to the Oracle Database Administrator's Guide.

3.24.1.14 Bug 15865494

When a PDB with an offline tablespace is plugged into a CDB, the tablespace cannot be brought online after the plug in is done. Currently, a warning is raised during plug in time to indicate this limitation.

Workaround: Before plugging in a PDB make sure there are no offline tablespaces. If there are any offline tablespaces make sure they are brought online before doing the plug in.

3.24.1.15 Bug 14246380

Using the ALTER SESSION SET CONTAINER statement in the definer's rights procedures can result in some session states not getting set up correctly.

Workaround: Use the DBMS_SQL package to execute a SQL statement in a given container, or use the ALTER SESSION SET CONTAINER statement outside of the definer's rights procedures.

3.24.1.16 Bug 14058913

If the primary database is a CDB and the control file is re-created such that it does not have all of the data file names, it will crash or interrupt the recovery process on the standby databases.

Workaround: Take the following steps on the primary database:

1. Shut down the primary database.

2. Delete the control file.

3. Start the primary database without mounting the database.

4. Re-create the control file including the missing data files.

5. Execute database recovery for the missing files on the primary database and open the database and the PDBs.

Then, continue with the following steps on the standby database:

1. Shut down the standby database.

2. On the primary database, create a new standby control file.

3. Replace the previous standby control file with the newly created standby control file.

4. Mount the standby database.

5. Create the standby redo logs.

6. Open the standby database and all of its PDBs.

7. Start the managed recovery process (MRP).

3.24.2.1 Bug 15845247

If you change the MAX_STRING_SIZE initialization parameter while creating a database using DBCA with a template that includes data files, the database creation fails with the following error:

ORA-14694: database must in UPGRADE mode to begin MAX_STRING_SIZE migration

Workaround: Leave the MAX_STRING_SIZE initialization parameter set to the default (STANDARD) or whatever it was set to in the template during DBCA. If you need to create a database with MAX_STRING_SIZE=EXTENDED, use the custom database template in DBCA. In a multitenant container database (CDB), the MAX_STRING_SIZE initialization parameter is a per-PDB parameter. The root CDB always uses STANDARD semantics, regardless of the parameter. You can change the MAX_STRING_SIZE initialization parameter for PDBs as needed after the CDB is created.

3.24.3.1 Bug 8644344

When running the deinstallation tool to deinstall the database, you will be prompted to expand the Oracle home and to select a component. If you select the top level component, Oracle Database Server, and do not select the Oracle home, OUI does not show the message to run the deinstall utility and proceeds with the deinstallation of the database.

Workaround: Run the deinstallation tool to deinstall the Oracle home.

3.24.4.1 Bug 16714677

When using Oracle Application Express Application Migration (Application Migration) with Java Database Connectivity (JDBC), the standard parameter markers (?) are converted to Oracle style in the format ":<n>". That is, each consecutive "?" is converted to ":1", ":2", and so on, before the query goes to the server for translation. However, due to an existing bug (also reference Bug 16182805), the conversion is not working reliably and converts the markers to the format " :b<n> " thus causing unreliable behavior. This incorrect format can cause inconsistent behavior during query translation. Unfortunately, this inconsistent behavior is documented in the Oracle Database SQL Translation and Migration Guide.

Workaround: The incorrectly documented behavior will be fixed in a future release. Any translation with parameters working in Oracle Database 12c is a coincidence and is not consistent.

3.24.4.2 Bug 16182805

When using Oracle Application Express Application Migration (Application Migration) with Java Database Connectivity (JDBC) driver, binds will not be translated correctly. This impacts the Oracle Sybase Translator which cannot work with bind variables in a JDBC program but is not limited to the Oracle Sybase Translator. Any translator relying on binds may be impacted.

Workaround: None.

3.24.4.3 Bug 16063217

Several changes were made to local transaction processing for compliance with the JDBC spec 4.1. If setAutoCommit(true) is called and a local transaction exists, the transaction is automatically committed (in earlier releases, no action was taken). If commit or rollback is called while in AUTOCOMMIT mode, the driver will throw an exception (again, in earlier releases, no action was taken). It is possible that your application will have these situations and it may be difficult to immediately change the software.

Workaround: By adding the -Doracle.jdbc.autoCommitSpecCompliant=false system property to the command line, the old behavior of no action will be preserved.

3.24.4.4 Bug 16062511

To use Oracle WebLogic Server (WLS) Active GridLink 10.3.6 or 12.1.1 with the new Oracle Database 12c Release 1 (12.1) driver, the -Doracle.ucp.PreWLS1212Compatible=true system property must be specified for compatibility.

Workaround: None.

3.24.4.5 Bug 15874495

Transaction Guard and Application Continuity do not support global (XA) transactions.

Workaround: None.

3.24.4.6 Bug 14069709

Oracle proxy authentication does not work with Application Continuity.

Workaround: None.

3.24.5.1 Bug 16891982

When upgrading Oracle Grid Infrastructure from 10.2.x.x to Oracle Database 12c Release 1 (12.1) and when the 10.2.x.x Oracle ASM home owner is different than the 12.1 Oracle Grid Infrastructure home owner, the upgrade will fail.

Workaround: Take the following steps:

1. After the failure, exit the Oracle Universal Installer (OUI).

2. As the Oracle Grid Infrastructure user, set the ORACLE_BASE environment variable to the Oracle base of the 12.1 Oracle Grid Infrastructure home.

3. Run the following command to finish the Oracle ASM upgrade:

   GRIDHOME/bin/asmca -silent -upgradeASM
   

3.24.5.2 Bug 16227068

When creating a disk group for use in Oracle Flex ASM mode that will use Oracle ASM Dynamic Volume Manager (Oracle ADVM) volumes, COMPATIBLE.ASM must be set to either 12.1 or 12.1.0.0.0 but not 12.1.0.1.0. Oracle ADVM volumes will not be able to be used in Oracle Flex ASM mode if COMPATIBLE.ASM is set to 12.1.0.1.0 during disk group creation. This issue does not exist if COMPATIBLE.ASM is set to 12.1.0.1.0 after disk group creation.

Workaround: Set COMPATIBLE.ASM to either 12.1 or 12.1.0.0.0.

3.24.5.3 Bug 16094980

After installing an Oracle Flex ASM cluster or upon completion of the conversion process from a classic Oracle ASM cluster to an Oracle Flex ASM cluster, enabling an Oracle ADVM volume experiences a timeout when the Cluster Ready Services (CRS) stack starts up in Oracle Flex ASM mode and the following error message may appear on the console:

CRS-5000: Expected resource ora.C3DBFRA.dg does not exist in agent process. 
For details refer to "(:CLSN00107:)" in   
"/TB/12.1.0/grid/log/hi07-3d/agent/crsd/orarootagent_root/orarootagent_root.log". 

Workaround: This condition is benign and may be ignored. Enabling the volume should succeed within a few minutes and the CRS volume resource will transition to an ONLINE state.

3.24.5.4 Bug 13503952

Starting in Oracle Database 12c, the default discovery string for Oracle ASM has changed to /dev/sd* on Linux.

Workaround: Explicitly set the discovery string appropriate for your system, in order to continue to use the /dev/raw/* disks.

3.24.5.5 Bug 12332603

Oracle Automatic Storage Management (Oracle ASM) loses the rolling migration state if Cluster Ready Services (CRS) shuts down on all nodes.

Workaround:  Consider the following scenario of 4 nodes (node1, node2, node3, and node4) that are at release 11.2.0.2 and being upgraded to Oracle Database release 12.1.0.1:

• node1 and node2 are upgraded to 12.1.0.1 and running.

• node3 and node4 are still at 11.2.0.2 and running.

Now consider that there is an outage where all CRS stacks are down which leaves the cluster in a heterogeneous state (that is, two nodes at 11.2.0.2 and two nodes at 12.1.0.1).

To proceed with the upgrade, only nodes at 11.2.0.2 (that is, node3 and node4 or both) should be started and the following command needs to be executed on the Oracle ASM instance on node3 and node4 before starting any 12.1.0.1 node:

ALTER SYSTEM START ROLLING MIGRATION TO '12.1.0.1'

Continue the upgrade procedure as already documented from this point forward.

Note that before executing the preceding step to bring the Oracle ASM cluster back into rolling migration, you cannot start two nodes of different versions in the cluster. If you do so, one of the Oracle ASM versions fail with either the ORA-15153 or ORA-15163 error message.

3.24.5.6 Bug 9181675

The asmgidwrap script needs to be called if you are creating a database manually on Oracle ASM to avoid a permission error.

Workaround: For a role-separated installation (that is, there is a different user and group for grid and RDBMS), use DBCA to create the database that automatically calls asmgidwrap script while creating a database on Oracle ASM. If you choose to create a database manually, the script needs to be called explicitly so the proper group can be set to avoid a permission error.

3.24.6.1 Bug 9683229

Oracle ADVM does not support mounting ext3 or ext4 file systems over Oracle ADVM with the mount barrier option enabled. The mount barrier option is enabled by default on SLES11.

Workaround:  Mount the ext3 or ext4 file system with -o barrier=1. For example:

mount -o barrier=0 /dev/asm/myvol-131 /mnt

3.24.7.1 Bug 16300127

Inactive volumes may prevent the Oracle Grid Infrastructure upgrade from completing.

Workaround:  If Oracle Universal Installer (OUI) reports a failure during the grid upgrade, take the following steps:

1. Ensure that all Oracle Automatic Storage Management Cluster File System (Oracle ACFS) file system resources are stopped by using crsctl stat res -t -w "TYPE == ora.acfs.type and crsctl stop res <resource name> -f.

2. Ensure that all volumes are enabled by using SQL or ASMCA to enable them.

3. Rerun the crs\config\gridconfig.bat script.

4. Continue the upgrade from OUI.

3.24.7.2 Bug 16170117

When upgrading from 11.2.x.x to Oracle Database 12c Release 1 (12.1), the rootupgrade script may fail to stop the 11.2 Cluster Ready Services (CRS) stack on a node if the Oracle ACFS resource was not stopped first.

Workaround:  Reboot one of 11.2.x.x nodes and then rerun the rootupgrade script.

3.24.7.3 Bug 16089462

During an upgrade from Oracle Database 11g to Oracle Database 12c, the Oracle Automatic Storage Management Cluster File System (Oracle ACFS) registry CRS resource is replaced with individual CRS Oracle ACFS file system resources corresponding with each registered Oracle ACFS file system.

As a part of this process, any Oracle ASM Dynamic Volume Manager (Oracle ADVM) volumes in the Oracle ACFS registry which no longer exist or are offline during the upgrade will result in errors during the conversion. For example:

PRCA-1089 : Unable to retrieve volume and disk group for volume device path <path>

Workaround:  This situation can be avoided by using the acfsutil registry -d command to remove volumes and mount points from the Oracle ACFS registry that are no longer valid.

Alternatively, if a volume cannot be enabled as part of the upgrade process, the upgrade will succeed but Oracle ACFS file systems associated with offline volumes will need to be manually re-added to the CRS namespace. This can be done with the following steps:

1. Enable the volume using SQL, Oracle ASM Configuration Assistant (ASMCA), or the srvctl start volume command.

2. Use the acfsutil registry -a command or the srvctl add filesystem command to re-add the Oracle ACFS file system to the CRS namespace.

3.24.7.4 Bug 16040501

When enabling Oracle ACFS plug-in for the first time on a particular node, the acfsutil plugin enable command may fail with the following error:

ACFS-3613: Unable to write plug-in config file.

Workaround:  List the contents of the Oracle ACFS root directory and then reissue the acfsutil plugin enable command.

3.24.7.5 Bug 14739783

In the rare event of a transparent high availability failure where dropping the volume does not trigger the removal of its corresponding volume resource, the resource can be removed using the following command:

srvctl remove volume

Workaround:  Remove the resource.

3.24.7.6 Bug 14513630

The High Availability NFS for Grid Home Clusters (HANFS) feature does not support IPv6 addresses in this release.

Workaround:  Configure the High Availability Virtual Internet Protocol (HAVIP) using a name that resolves only to an IPv4 address.

3.24.7.7 Bug 14139048

When modifying a registered file system using acfsutil registry and if there are exports or databases dependent on the file system, the file system will still be modified. For instance, this could result in a non-supported configuration if you have changed a file system from a cluster-wide system to a 'node local' file system.

Workaround:  None.

3.24.7.8 Bug 10069735

In a cluster with a password-protected key store, when an Oracle ACFS file system using encryption is mounted through the Oracle ACFS mount registry, the administrator is not prompted to enter the key store password. Although the process of mounting the file system succeeds, not all information required for Oracle ACFS encryption to work correctly is made available to the file system. In this case, encryption is not operational on this file system and any encrypted files in the file system are not available for read or write.

Workaround:  In a cluster with a password-protected key store, do not use the Oracle ACFS mount registry for mounting any file systems that are using encryption. If some file systems are already mounted through the Oracle ACFS mount registry, unmount them and remove these file systems from the mount registry to avoid possible unavailability of encrypted data in the future. Then, remount these file systems without using the Oracle ACFS mount registry, providing the correct password when requested. Alternatively, if you want to continue to use the mount registry with encrypted file systems, migrate the encryption keystore using acfsutil keystore migrate command to the type that does not require a password during file system mounting.

3.24.8.1 Bug 16992360

Resource Manager changed the way CPU resources are managed for CDB or PDB database deployments in a manner that was incompatible with Oracle Database QoS Management 11.2 plan and models. These changes resulted in the need for two plans and different resource modeling with associated workload validation. These models need to be developed, tested, and calibrated on production Resource Manager code. Therefore, in this initial release, Oracle Database QoS Management is only able to measure and monitor CDB or PDB database deployments and cannot make recommendations to help CDB or PDB performance classes that are violating their performance objectives.

Workaround: None.

3.24.8.2 Bug 12792222

This bug applies to recommendations for CPU resources managed by Oracle Database QoS Management. If the number of configured CPUs for all instances on a server is less than the number of physical CPUs for that server, then the nonallocated, or "free", CPUs are not detected by Oracle Database QoS Management and no recommendation is made to increase the number of configured CPUs. Only those "slices" that host databases are considered as donors for the target slice. Adding one of the non-allocated CPUs should be the first-ranked Move CPU action.

Workaround: Make sure the sum of CPU counts configured for each database instance on each server is the same as the number of physical CPUs.

3.24.8.3 Bug 10019209

This bug applies to platforms that support the Cluster Health Monitor (CHM). If an Oracle Clusterware-managed database service is in a stopped but not disabled state, it will be started by Oracle Database QoS Management if the server hosting that service is not detected to be in a memory overcommitted state. If memory is overcommitted, then all enabled services will be stopped even if they were manually started. The desired behavior is to only start services on the transition from a memory overcommitted state (red) to a normal state (green). If a service is manually started when the server is in the red state, that service should not be shut down.

Workaround: Stop and disable services that you want to remain in the stopped state or disable Oracle Database QoS Management from the Oracle Enterprise Manager Console.

3.24.9.1 Bug 16914379

Upgrading the Oracle Grid Infrastructure from Oracle Database 11g Release 2 (11.2.0.4) to Oracle Database 12c Release 1 (12.1) fails due to Oracle ACFS or Oracle ADVM resources being unable to stop.

Workaround: Manually stop the resources using SRVCTL or CRSCTL, and retry the upgrade.

3.24.9.2 Bug 16904822

In an Oracle Flex ASM cluster configuration with a pre-11.2 database setting, the pre-11.2 database's CLUSTER_INTERCONNECTS initialization parameter uses a public network instead of a private network if the Oracle Cluster Synchronization Services (CSS) private network and Oracle ASM network are classified to be the same network.

Workaround: Select a different network for the cluster interconnect and Oracle ASM.

3.24.9.3 Bug 16899080 and Bug 14693336

The Oracle Cluster Synchronization Services (CSS) or Oracle ASM instances cannot rejoin the cluster after a failure in all of the private networks.

As a prerequisite, ensure that at least one private network has been restored on the failed node.

Check the cluster state. The follow commands can be used to determine if instances cannot rejoin the cluster:

crsctl stat res ora.cssd -init
crsctl stat res ora.asm -init
crsctl check cluster -all

Workaround: Reboot the rejoining node. If rebooting the node is not acceptable, then the following sequence may be followed:

1. Stop the clusterware on the node that cannot rejoin the cluster using the following command:

   crsctl stop crs –f
   

2. Stop any remaining Highly Available IP Addresses (HAIP) if HAIP is in use on the system. HAIP addresses are in the 169.254.0.0 subnet by default. Use operating system-specific commands to stop the interfaces. For example, on Linux use:

   ifconfig eth0:1 down
   

   On Solaris use:

   ifconfig net2:1 unplumb
   

   On AIX use:

   ifconfig en4 169.254.180.68 delete
   

3. Check that at least 10 minutes have passed from the point at which it was noticed that the node failed to join the cluster. Restart the clusterware using the following command:

   crsctl start crs
   

4. If the node still does not rejoin the cluster, reboot the node.

3.24.9.4 Bug 16734478

If you upgrade an 11.2 Oracle Clusterware having a large number of services to Oracle Database 12c Release 1 (12.1), the rootupgrade.sh script that runs on the last cluster node aborts because the ora.crsd startup initialized but failed to complete during the upgrade.

Workaround: Rerun the rootupgrade.sh script on the last node.

3.24.9.5 Bug 16684285

Upgrading from 11.2.x.x to Oracle Database 12c, when there are more than 1,000 database service resources registered, the root script will fail on the first node while trying to upgrade the Server Management (SRVM) model. This problem may prevent the Cluster Ready Services (CRS) upgrade from 11.2.x.x to Oracle Database 12c from completing when there are more than 1,000 database service resources.

Workaround: Patch the 11.2 home to the version with the fix or remove database service resources, complete the upgrade, and re-register the database service resources.

3.24.9.6 Bug 16543421

Oracle Universal Installer returns the INS-41112 error when /usr/sbin/ping6 does not have setuid set.

Workaround: Run the following command as root to set the setuid and rerun the installer:

chmod +s /usr/sbin/ping6

3.24.9.7 Bug 16539062

In some cases in an Oracle Restart environment, the Oracle Clusterware stack generates a connection to the Oracle ASM instance every second which results in producing many audit files under Grid_Home/rdbms/audit that may fill up the disk over time.

Workaround: The following command will immediately stop these audit files from being generated:

/u01/app/oracle/product/12.1.0/grid/bin/crsctl modify resource ora.asm 
-attr START_DEPENDENCIES="hard(ora.cssd)" 

3.24.9.8 Bug 16468516

In an environment with 5 or more voting files, if a node is shut down abruptly, it is possible that the Oracle Cluster Synchronization Services (CSS) daemon will write its last disk heartbeat to only a subset of voting files for that node. This would result in a situation where two voting files have different disk heartbeat timestamps for the node. In this situation, if you attempt to shut down an entire cluster and start another node (other than the one that was shut down abruptly) in EXCLUSIVE MODE, the EXCLUSIVE MODE start could fail as it might incorrectly perceive the abruptly shut down node to be up.

Workaround: Use the following steps to recover from this problem:

1. Either physically remove all of the voting files or make them offline.

2. Bring up the stack in EXCLUSIVE MODE.

3. Delete all the voting files from the configuration using the crsctl delete css votedisk command.

4. Add new voting files using the crsctl add css votedisk command.

The stack can now be brought up in Oracle Flex Cluster mode.

3.24.9.9 Bug 16435381

Creation of an 11.2-based database on a newly installed 12.1 Oracle Grid Infrastructure fails.

An upgraded Oracle Grid Infrastructure from a pre-12.1 release to 12.1 will not have a problem running an 11.2-based database.

Workaround: None.

3.24.9.10 Bug 16407903

Oracle resources for release 10.2 and release 11.1 Oracle RAC databases may not operate properly after upgrading Oracle Clusterware to Release 12.1.

Workaround: After installing Oracle Clusterware 12g Release 1 (12.1), contact Oracle Support Services to obtain the patches for the following bugs:

• 8373758 - TB-CMP: 11107 SERVICE CAN'T BE BROUGHT UP BY 11107 SRVCTL WHEN WITH 11.2 CRS

• 8441769 - TB_UD: INCORRECT SERVICE INFO REGISTER TO DB, UPGRADE CRS_HOME 11.1.0.7 -> 11.2

• 8406545 - TB-CMP: RESTART OF 11.2 HAS STACK FAILED TO BRING UP 11.1 ORACLE RAC INSTANCE

• 8262786 - TB-CMP: FAIL TO START 11106 DB INSTANCE WITH 11.2 CRS

3.24.9.11 Bug 16324841

When opening an instance of an Oracle Real Application Clusters (Oracle RAC) multitenant container database (CDB), the open may hang. This is because the pluggable databases (PDBs) were not opened uniformly across all instances.

Consider PDB1 of a CDB being opened on Instance 1 but not on Instance 2. Also, consider that there was a dead transaction that originated on Instance 2 that modified PDB1 sometime in the past. Since PDB1 is open on Instance 1, it is possible for processes to try and modify data already modified by that dead transaction. However since the undo segment belongs to Instance 2, Instance 1 cannot recover the dead transaction. This dead transaction, however, cannot be recovered on Instance 2 either since PDB1 is closed on Instance 2. This leads to the hang.

Workaround: Start PDB1 on all of the instances that are open.

3.24.9.12 Bug 16246278

In Oracle Grid Infrastructure release 12.1 on the Linux platform, the default Oracle ASM discovery string has been modified from /dev/raw/raw* to /dev/sd*. If the default Oracle ASM discovery string is being used for selecting the devices managed by Oracle ASM and if the existing character device does not map to their respective block devices which are identified by pattern /dev/sd*, then these devices will not be visible to Oracle ASM. This results in disk groups not being mounted after the upgrade causing the upgrade to fail. The Cluster Verification Utility (CVU) prerequisite check for these devices will also be skipped.

Workaround: Set the Oracle ASM disk discovery string by using the following command before the upgrade:

ALTER SYSTEM SET ASM_DISKSTRING=<ASM_disk_discovery_string> SCOPE=SPFILE;

3.24.9.13 Bug 16168869

In the IPv6 environment only, some SCAN virtual IP addresses (VIPs) are not registered in the Grid Naming Service (GNS).

Workaround: If GNS is running in the cluster, use the srvctl relocate gns -node <node_name> command to relocate GNS to the node on which SCAN VIP is running. This allows SCAN VIP to be registered to GNS. Afterward, GNS can be relocated to other nodes.

3.24.9.14 Bug 16095676

When you have provided the sudo or root user in the Oracle Universal Installer (OUI) GUI pages in order to use the Oracle Grid Infrastructure script automation feature, the root script runs using this automation feature but may fail with the following error:

Exception in thread "main" java.lang.NullPointerException

This happens because IPv6 addresses could not be resolved through one of the named servers defined in the /etc/resolve.conf file.

Workaround: Open the /etc/resolve.conf file and collect all of the named servers defined in this file. Usually, the following are defined:

nameserver x1.y1.z1.k1
nameserver x2.y2.z2.k2
nameserver x3.y3.z3.k3

For each of the named servers, run the following command to determine if the IP addresses are resolved:

nslookup -querytype=AAAA <IPv6_address> <named_server>

If you see error messages similar to the following, it means that the particular named server cannot resolve the IPv6 addresses:

;; connection timed out; trying next origin 

You need to remove or comment the named servers which cannot resolve the IPv6 addresses.

3.24.9.15 Bug 16095320

In a Grid Naming Service (GNS) enabled configuration, the client cannot connect to the database using the SCAN name. This is because the SCAN name in the REMOTE_LISTENER value of the database is not domain-qualified.

Workaround: Using SQL*Plus, manually modify REMOTE_LISTENER to be a domain-qualified SCAN name. For example:

SQL> ALTER SYSTEM SET REMOTE_LISTENER 
='myscan.mycluster.example.com:1521' SCOPE=BOTH; 

3.24.9.16 Bug 16088332

When upgrading Oracle Grid Infrastructure to 12.1, you may see following error message on output stream:

PRCR-1154 : Failed to create file output stream with file name: 
GI Home/cfgtoollogs/crsconfig/srvmcfg0.log 

Workaround: None. You can ignore this error message. The upgrade completes irrespective of this warning.

3.24.9.17 Bug 16083544

After upgrading the Oracle Grid Infrastructure from 11.2.0.3 to 12.1, the 11.2.0.3 database (administrator managed) does not come up.

Workaround: After the upgrade is finished, bounce Cluster Ready Services (CRS) by running crsctl stop crs followed by crsctl start crs.

3.24.9.18 Bug 16082183

The SCAN listener rejects service registration from nodes in the same cluster. The rejections are logged in the SCAN listener log file.

Workaround: Use srvctl modify scan_listener -update -invitednodes or srvctl modify scan_listener -update -invitedsubnets to specify the nodes or subnet information which the SCAN listener should accept for service registration.

3.24.9.19 Bug 16044275

When upgrading 11.1 CRS to 12c Oracle Grid Infrastructure for a cluster and after Oracle ASM Configuration Assistant (ASMCA) upgrades Oracle ASM, some database instances may not be started. This problem is not reflected in the Oracle Universal Installer as an error or warning.

Workaround: Manually check the database instance after upgrading to detect the state of the database instances. For example:

srvctl status database -db <db_name>

Then, manually start the database instance. For example:

srvctl start database -db <db_name>

3.24.9.20 Bug 16039546

The Grid Naming Service (GNS) integrity check may fail while running cluvfy commands if the GNS virtual IP address (VIP) is resolved to IPv4 and IPv6 addresses in a pure IPv6 environment.

Workaround: Query the IPv6 address of the GNS VIP. If the query is successful, you may ignore the GNS integrity failures.

3.24.9.21 Bug 16008999

The Grid Infrastructure Management Repository's (GIMR) security policy requires its users' password to be changed every 180 days. In this release, the GIMR clients do not perform this function and thus OLOGGERD and OCLUMON will log connection exceptions.

Workaround: A CRS administrator needs to run (manually or otherwise) the GRID_HOME/bin/mgmtca command from the node running the GIMR. That node can be found by running the GRID_HOME/bin/srvctl status mgmtdb command.

3.24.9.22 Bug 16003413

Interrupting or killing the installation script on the first cluster node may result in the following errors when the script is run again on the first node or is subsequently run on other cluster nodes:

CLSRSC-46: Error: '$ORA_CRS_HOME/gpnp/wallets/pa/cwallet.sso' does not exist 
CLSRSC-153: Could not set permissions on 
  '$ORA_CRS_HOME/gpnp/wallets/pa/cwallet.sso' 
CLSRSC-148: Errors occurred while setting GPnP wallets ownership/permissions 

These errors are caused by a missing copy of a non-essential file which is not detected when the existing configuration revalidated on subsequent runs.

Workaround: These errors may be safely ignored, the script will continue to run, and the product installation will not be affected.

However, if a clean run is desired, delete the files in following directories on all cluster nodes before the script restarts on the first node:

rm $ORA_CRS_HOME/gpnp/profiles/peer/* 
rm $ORA_CRS_HOME/gpnp/wallets/peer/* 

3.24.9.23 Bug 15894162

When multiple clusters use the same directory on a shared file system to place the Oracle Cluster Registry (OCR) and voting disks, installation and upgrade with mgmtdb fails.

Workaround: Use different storage locations for OCR or make sure that the mgmtdb option is disabled for all of the clusters except one.

3.24.9.24 Bug 15882236

When a node joins a cluster after a force upgrade, the database instances and services do not start automatically.

Workaround: Manually start the database instances and the services on this node using the SRVCTL located in that database home.

3.24.9.25 Bug 14634152

In a Chinese environment (that is, a locale other than zh_CN.utf8 or zh_TW.utf8), during the Oracle Grid Infrastructure installation, if you enable the "Use "root" user credential" option, the GUI will hang and the INS-32128 error is returned.

Workaround: There are two workarounds for this bug:

• Workaround 1: Set the locale to zh_CN.utf8 or zh_TW.utf8.

• Workaround 2: Do not enable this option. Manually execute root.sh.

3.24.9.26 Bug 14497206

Service resources for pre-11.2 releases may be OFFLINE after Oracle Grid Infrastructure is upgraded to release 12.1.

Workaround: Use srvctl start service -d <dbname> -s <srvname> -i <instname> to start the OFFLINE service resources manually.

3.24.9.27 Bug 13993872

On rare occasions, when upgrading Oracle Clusterware from 11.1.0.7.7 to 12.1, node reboot issues may be encountered while executing rootupgrade.sh, which only occurs in the following conditions:

• The 11.1.0.7.0 release was installed on the cluster and the clusterware was started on a node before patch 7483779 was applied (which is a BLR fix for Bug 7374972).

• A full clusterware restart has not occurred since patch 7483779 was applied. A full clusterware restart is a clusterware start after being down on all nodes in the cluster.

If the initial install was 11.1.0.7.1 (that is, 11.1.0.7 CRS Bundle 1 or later), patch 7374972 was included in the initial install, so this problem will not be observed.

Workaround: Perform a full restart of the clusterware on all nodes before upgrading to 12.1.

3.24.10.1 Bug 19559050

The Oracle OLAP option is only available with Oracle Database Enterprise Edition. An error occurs if you attempt to use Oracle OLAP with Oracle Database Standard Edition (SE). For example, if you try to export using Oracle OLAP with SE, you see the following errors:

EXP-00008: ORACLE error 29280 encountered
ORA-29280: invalid directory path
ORA-06512: at "SYS.UTL_FILE", line 41
ORA-06512: at "SYS.UTL_FILE", line 478
ORA-06512: at "SYS.DBMS_AW_EXP", line 89
ORA-06512: at "SYS.DBMS_AW_EXP", line 1177
ORA-06512: at line 1
EXP-00085: The previous problem occurred when calling
SYS.DBMS_AW_EXP.instance_extended_info_exp for object 85676

Workaround: Use Oracle OLAP with Oracle Database Enterprise Edition.

3.24.10.2 Bug 18718327

For pluggable database (PDB) maintenance operations, parallel tasks need to be allocated on all active instances to perform the operation on each instance regardless of the setting of the PARALLEL_MAX_SERVERS initialization parameter. If an instance had PARALLEL_MAX_SERVERS=0 set, the parallel task was not allocated on the instance and the operation was not performed there.

Workaround: Do not set the value of the PARALLEL_MAX_SERVERS initialization parameter to 0. Not setting the PARALLEL_MAX_SERVERS initialization parameter at all is sufficient.

3.24.10.3 Bug 16933305

Adding a nullable column with a default value and then later unsetting the default value in an Oracle Database 12c Release 1 (12.1) environment will not unset the default (back to NULL) despite the data dictionary reflecting the change to a default value of NULL. The statements that are impacted are ALTER TABLE x ADD (y NUMBER DEFAULT 99) followed by a subsequent ALTER TABLE x MODIFY (y DEFAULT NULL) where y is a nullable column.

Workaround: Unset the default by using the ALTER TABLE x MODIFY (y DEFAULT TRIM('')) statement which has the same semantic result as unsetting the default.

3.24.10.4 Bug 16916434

If Oracle RAC database files are located on Network File Storage (NFS) or Oracle Cluster File System 2 (OCFS2), set the FILESYSTEMIO_OPTIONS=DIRECTIO or FILESYSTEMIO_OPTIONS=SETALL initialization parameter to avoid data corruption. This issue does not apply to Direct NFS. A patch will be available in the future. Monitor Bug 16865261 for the latest status and patch availability.

Workaround: Set the FILESYSTEMIO_OPTIONS initialization parameter to either the DIRECTIO or SETALL value. Note that this setting is not required if the database files are located on NFS and Direct NFS is enabled.

3.24.10.5 Bug 16873173

It is possible for an index fast full scan of a prefix compressed index to fail with ORA-600[6033] error.

Workaround: Retry the query or force an alternative access path with an appropriate hint.

3.24.10.6 Bug 16788520 and Bug 15968058

If shared server is enabled, queries on object link views in a PDB may crash. Object link views are all Oracle supplied and are mostly DBA_HIST views. A full list can be found using the following command:

SELECT OWNER, OBJECT_NAME FROM ALL_OBJECTS WHERE SHARING='OBJECT LINK' AND OBJECT_TYPE='VIEW'

Workaround: Disable shared server when querying these views.

3.24.10.7 Bug 16571451

SQL plan directives are not used when the OPTIMIZER_DYNAMIC_SAMPLING initialization parameter is set to the default value of 2.

Workaround: Set the OPTIMIZER_DYNAMIC_SAMPLING initialization parameter to a value greater than 2.

3.24.10.8 Bug 16527374

If a common user queries certain Oracle-supplied views in the root and the common user is restricted by the CONTAINER_DATA attribute to see only the data from certain containers, the query may result in an ORA-7445 error.

Workaround: Switch to the root and set the CONTAINER_DATA attribute of the common user to be ALL.

Execute the following ALTER USER statement:

ALTER USER <common_user> SET CONTAINER_DATA=ALL CONTAINER=CURRENT;

3.24.10.9 Bug 16525931

Oracle RAC Load Balancing (RLB) events for pluggable database (PDB) services are missing from the Oracle Notification Service (ONS) event stream. Code that relies on these events will not receive them.

Workaround: None.

3.24.10.10 Bug 16309162

When the DBMS_AUDIT_MGMT.CONTAINER_ALL value is used in DBMS_AUDIT_MGMT routines that contain the CONTAINER parameter, the execution fails with the following error:

ORA-46273: DBMS_AUDIT_MGMT operation failed in one of the PDB

Workaround: Run the required routine after connecting to the respective PDB and use the DBMS_AUDIT_MGMT.CONTAINER_CURRENT value for the CONTAINER parameter.

3.24.10.11 Bug 16213407

Pre-11.2 Oracle instance resources may not be started automatically when the 12.1 Oracle Grid Infrastructure stack together with an Oracle Flex ASM configuration starts.

Workaround: Manually start the pre-11.2 Oracle instance resource.

3.24.10.12 Bug 16089104

When converting a cluster from classic Oracle ASM to Oracle Flex ASM, one or more occurrences of the following error message sequences are observed in the output of the script that is executed as root by the user:

CRS-2672: Attempting to start 'ora.proxy_advm' on '<node-name>'^M 
CRS-5017: The resource action "ora.proxy_advm start" encountered the   
following error: ^M 
ORA-03113: end-of-file on communication channel^M 
Process ID: 0^M 
Session ID: 0 Serial number: 0^M

For details, refer to (:CLSN00107:) in <CRS-Home>/log/<node-name>/agent/crsd/oraagent_crsusr/oraagent_crsusr.log.

Workaround: These errors may be ignored. At the end of the conversion, the ora.proxy_advm correctly goes to an ONLINE state on all of the nodes.

3.24.10.13 Bug 16047938

Concurrent UNION ALL is automatically invoked for qualifying statements only when the UNION ALL statement is in a subsequent SELECT statement. For example, the following command will enable execution of all branches concurrently:

SELECT * FROM (SELECT FROM ... UNION ALL ... UNION ALL)

However, the exact same UNION ALL statement not executed as a subsequent SELECT statement will not.

Workaround: Either embed the UNION ALL construct as a subsequent SELECT statement or use the following statement to disable legacy code constraints:

ALTER SESSION SET "_fix_control"='6748058:0';

3.24.10.14 Bug 16027608

If you have an AL32UTF8 database, and if SQL*Loader is used with external tables as the load method and the table name contains non-ASCII characters, SQL*Loader may exit without an error and not load any data into the target database. This can occur either:

• With a control file if external_table=execute is specified.

• When using express mode if the default load method of external tables is used or is forced using the external_table=execute command line parameter.

Workaround: None.

3.24.10.15 Bug 15971421

The soft limit of nproc is not adjusted at runtime by the database. As a result, if that limit is reached, the database may become unstable since it will fail to fork additional processes.

Workaround: Ensure that the soft limit for nproc in /etc/security/limits.conf is set high enough to accommodate the maximum number of concurrent threads on the system for the given workload. If in doubt, set it to the hard limit. For example:

oracle  soft    nproc   16384 
oracle  hard    nproc   16384 

3.24.10.16 Bug 15881592

While enabling Oracle Flex ASM, there is a requirement to do a rolling restart of all the nodes in the cluster. When a node with management database instance (or database instances managed by Oracle Restart) is restarted, that instance might get relocated to another node which has not been restarted yet. In this case, the database instance will fail to start with the following error:

ORA-15343: Feature Flex ASM is not enabled

This will happen until the database instance is started on a node that has been restarted. When that happens, the database instance will behave normally.

Workaround: None. Once all the nodes are restarted, the database instance will behave normally.

3.24.10.17 Bug 15851422

The PQ_CONCURRENT_UNION hint currently does not work as a statement-level hint but requires the explicit specification of a query block. For example, the following command works and does concurrent UNION ALL processing:

SELECT /*+ PQ_CONCURRENT_UNION (@SET$1) */ ... UNION ALL ... UNION ALL ...

However, the following command does not work:

SELECT /*+ PQ_CONCURRENT_UNION */ ... UNION ALL ... UNION ALL ... UNION ALL ...

Refer to the Oracle Database SQL Tuning Guide for information on how to identify and name a query block.

Workaround: Explicitly specify a query block.

3.24.10.18 Bug 14851837

Tables containing TIMESTAMP WITH LOCAL TIME ZONE data cannot be moved between databases using transportable tablespace technology when the time zones of the source and target databases are different. Each affected table will be flagged during import with the following error:

ORA-39360, Table "<owner>"."<table name>" skipped due to transportable import and TSLTZ issues. 

Workaround: Convert the target database to the same time zone as the source database or move the affected tables with the conventional Data Pump Export and Import.

3.24.10.19 Bug 14734707

The Oracle interval data type, INTERVAL YEAR TO MONTH, does not work properly with the Ethiopian calendar across all components such as SQL, PL/SQL, OCI, Pro*C, and so on.

Workaround: None.

3.24.10.20 Bug 14621745

Certain errors raised while using the ALTER TABLE MOVE ONLINE statement can be ignored.

Workaround: None.

3.24.10.21 Bug 14558315

Using indexes for indexed partitions or a full table scan for partitions with no indexing is necessary to leverage partial indexing to its fullest extent. Oracle Database 12c Release 1 (12.1) currently does not support table expansion for interval partitioned tables, the functionality to internally rewrite a statement into a UNION ALL to leverage indexes whenever possible.

Workaround: Either manually rewrite a statement into a UNION ALL operator to separate partitions with and without indexing or to disable the interval partition functionality.

3.24.10.22 Bug 13790316

Starting Oracle Database 12c Release 1 (12.1), temporary LOBs sized up to 256K live in program global area (PGA) memory. This can lead to increased PGA memory consumption. Some workloads may encounter the ORA-4030 error depending on the number of temporary LOBs created.

Workaround: Set event 32761, level 16 to turn off in-memory temporary LOBs. Setting this event spills temporary LOBs to temporary segments on disk. Even though this brings the memory consumption to pre-12.1 levels, users will not see the performance benefits of in-memory temporary LOBs.

3.24.11.1 Bug 16765611

During an upgrade from Oracle Database 11g Release 2 (11.2.0.4) to Oracle Database 12c Release 1 (12.1) with Oracle Database Vault already installed in 11.2.0.4, you get the following two ORA-00001 errors in the upgrade log:

ORA-00001: unique constraint (DVSYS.CODE$_PK1) violated
ORA-00001: unique constraint (DVSYS.CODE_T$_UK1) violated

Also, you will see the same errors from the post-upgrade status utility (utlusts.sql).

Workaround: These errors are not harmful and can safely be ignored.

3.24.12.1 Bug 16063508

LogMiner background processes may encounter ORA-54013 error after downgrading the database to a pre-12.1 release.

Workaround: Modify the table to remove the virtual column. If it was added using the DBMS_STATS.CREATE_EXTENDED_STATS() function, then the DBMS_STATS.DROP_EXTENDED_STATS() function can be used to remove the virtual column. If it is supporting a descending index, then drop the descending index.

3.24.12.2 Bug 16055604

Piecewise LOB updates on typed index-organized tables (IOT) are not replicated on a logical standby database. SQL Apply will stop with ORA-1403 when it encounters such a change in the redo stream.

Workaround: Use the DBMS_LOGSTDBY.SKIP procedure at the logical standby to skip the table from being replicated.

3.24.12.3 Bug 15847339

Supplemental logging is unsupported for updates of Abstract Data Type (ADT) attributes with a bind value of more than 4,000 bytes.

Workaround: To update ADT attributes that need replicating, use direct bind only if the bound value is smaller than 4,000 bytes (32,767 bytes if max_string_size=extended). To bind variables with values beyond this limit, either bind a LOB data type or update the entire ADT.

3.24.12.4 Bug 14675706

SQL Apply may stop with the following error while an Oracle Data Guard configuration is in the middle of a rolling change operation initiated by the DBMS_ROLLING package:

ORA-00600: internal error code, arguments: [kwqicarsm: rs obj #], []

Workaround: The DBMS_RULE_ADM package is not supported in the context of the DBMS_ROLLING package and should not be used to create or alter rules for Oracle Advanced Queuing (AQ) subscribers at the primary database while DBMS_ROLLING is active.

Use DBMS_ROLLING.ROLLBACK to roll back the rolling change operation in the Oracle Data Guard configuration. Retry the rolling change operation once you have ensured that DBAs or applications are not going to add or change rules associated with AQ subscribers for the duration of DBMS_ROLLING operations.

3.24.12.5 Bug 14530359

SQL Apply does not support replication of a table with SYS.ANYDATA column, if the SYS.ANYDATA column contains multibyte characters.

Workaround: None.

3.24.12.6 Bug 14538279

SQL Apply aborts with the following error on a table with a Securefile LOB column:

ORA-600 [curr loc with no gather] [0x2B6D75918]

Workaround: Skip the table using DBMS_LOGSTDBY.SKIP and restart SQL Apply.

3.24.12.7 Bug 14529358

SQL Apply does not support replication of a table with Abstract Data Type (ADT) columns when the type names (or the associated attributes) contain multibyte characters.

Workaround: None.

3.24.12.8 Bug 14400003

DBMS_LOGSTDBY.INSTANTIATE_TABLE does not support tables with invisible columns.

Workaround: None.

3.24.12.9 Bug 12911818

For Oracle Text, indexes created with the FILTER BY or ORDER BY clause of the SQL CREATE INDEX statement, DML statements on the columns specified in the FILTER BY or ORDER BY clause are not replicated at the logical standby. However, the indexes may not be properly maintained at the logical standby.

Workaround: Synchronize the indexes manually at the logical standby database.

3.24.13.1 Bug 16783586

The Enterprise Manager Agent should use a dedicated server connection to connect to a CDB for monitoring.

Workaround: This can be configured by setting (SERVER=dedicated) in the CONNECT_DATA of the connect string for the Enterprise Manager Agent connection.

3.24.13.2 Bug 15990064

Oracle home is not updated in Enterprise Manager after a database upgrade to Oracle Database 12g Release 1 (12.1) with an older version of Enterprise Manager.

Workaround: Upgrade Enterprise Manager to 12cPS1 or above with Database Plugin 12.1.0.3.0 or above to support target upgrade OOB on database creation or deletion. And, upgrade or modify the target properties manually by editing the monitoring property (Oracle_home) for the database and Listener target and change the Oracle home value from the old database home to the new 12.1 database home.

3.24.13.3 Bug 15990057

Database target addition, deletion, and upgrade are not supported with Enterprise Manager releases prior to 12cPS1 because these older releases of Enterprise Manager do not support web services.

Workaround: Upgrade Enterprise Manager to release 12cPS1 or above with Database Plugin 12.1.0.3.0 or above to support target addition and modification OOB on database creation, deletion and upgrade.

3.24.13.4 Bug 13029141

After parallel DML is done on a SecureFile LOB column from a relational table, a parallel SELECT query accessing the same SecureFile LOB column data intermittently shows the ORA-08103 error message.

Workaround: None.

3.24.14.1 Bug 9917299

If the database is installed using the seed provided in the installation kit, and the OLAP option is not selected, then either at the end of the installation or some time later, the OLAP Analytic Workspace and OLAP API components will be reported as invalid.

This will not affect the running of the instance in any way, other than the error messages.

Workaround: Do one of the following as a workaround:

• Ignore the error.

• Enable OLAP (or the offending option).

• Create and use your own seed database that does not include OLAP.

3.24.15.1 Bug 16233295

When upgrading from 11.2.x.x to 12.1.0.1, if any customized listener configuration is set (such as a remote listener), Database Upgrade Assistant (DBUA) will set it back to the default.

Workaround: Execute ALTER SYSTEM SET REMOTE_LISTENER=<user_remote_listener_setting>, ... SCOPE=BOTH from SQL*Plus after the database upgrade.

3.24.15.2 Bug 16219867

When an Oracle RAC policy-managed database is created by the Database Configuration Assistant (DBCA), the EM Express URL displayed by the DBCA summary may not work. This can happen if the local node is not part of the server pool hosting the database. This issue happens only for policy-managed Oracle RAC databases.

EM Express is configured and accessible from nodes where database instances are running. It can also be accessed using the scan name.

Workaround: Use the scan name for the host name in the EM Express URL. For example, consider the following EM Express URL (where racnode1 is the node name):

https://racnode1.oracle.com:5500/em 

You can specify the following (where scan1 is the scan name of the cluster):

https://scan1.oracle.com:5500/em 

3.24.15.3 Bug 16102371

The Oracle Enterprise Manager Database Express (EM Express) URL may not work on an Oracle RAC node.

Workaround: This bug may be encountered during Oracle RAC installation. If the EM Express URL does not work on an Oracle RAC node, restart the Node Listener on that machine.

3.24.15.4 Bug 16094984

If an Oracle RAC Node Listener rejects the service registration from an instance on the same node, the rejection gets logged in the listener log file with TNS-01182 error.

Workaround: Use lsnrctl RELOAD command from the operating system or restart the listener.

3.24.15.5 Bug 16089419

Using lsnrctl on Grid home cannot connect to the Node Listener even it is running.

Workaround: Use the srvctl getenv listener command to check the setting of TNS_ADMIN of the corresponding listener resource. Usually, it is similar to the following:

<GIHOME>/network/admin/<listener_user_id>

Then, set the TNS_ADMIN environment to that same value and rerun lsnrctl after which lsnrctl should able to connect to that listener.

3.24.15.6 Bug 16033089

When installing 12.1 Oracle Grid Infrastructure stack in IPv6 environments, there may be an issue in detecting the IPv6 network interfaces. This happens if there are no IPv4 interfaces or network interfaces which support both IPv4 or IPv6. This may cause the installer to fail to deploy Oracle Grid Infrastructure.

Workaround: Make sure to use network interfaces which support both IPv4 and IPv6 or activate a virtual IPv4 network interface. You can do this by performing the following command where net0 is the name of IPv4 network interface:

/sbin/ifconfig net0 up

3.24.15.7 Bug 16020794

When issuing the command srvctl stop listener -force to stop an Oracle ASM listener on a node where an Oracle ASM instance is running, a failure may be encountered with the following message:

PRCR-1014 : Failed to stop resource ora.ASMNET1LSNR_ASM.lsnr 
PRCR-1065 : Failed to stop resource ora.ASMNET1LSNR_ASM.lsnr 
Cannot communicate with crsd 

Workaround: If, as part of removing an Oracle ASM network, it is no longer required to run an Oracle ASM listener on the specific subnet. Use the following command:

srvctl update listener -listener <listener_name>  -asm -remove

This command stops the Oracle ASM listener on all nodes it is running and also removes the Oracle ASM listener configuration.

3.24.15.8 Bug 14842114

Pre-11.2 listeners fail to start after an Oracle Grid Infrastructure downgrade.

Workaround: Manually backup listener.ora from $db_home/network/admin before you upgrade and restore it after you downgrade.

3.24.15.9 Bug 14742343

After CRS is restarted, disk group resources may be OFFLINE even though they were mounted in Oracle ASM.

Workaround: Modify the disk group resource AUTO_START attribute to always. For example:

crsctl modify resource ora.<dgname>.dg  -attr AUTO_START=always 

Alternatively, you can accomplish the same thing by dismounting and mounting the disk group in Oracle ASM using SQLPLUS. For example:

ALTER DISKGROUP <dgname> DISMOUNT; 
ALTER DISKGROUP <dgname> MOUNT;

3.24.15.10 Bug 14575666

In 12.1, the default value for the SQLNET.ALLOWED_LOGON_VERSION parameter has been updated to 11. This means that database clients using pre-11g JDBC thin drivers cannot authenticate to 12.1 database servers unless the SQLNET.ALLOWED_LOGON_VERSION parameter is set to the old default of 8.

This will cause a 10.2.0.5 Oracle RAC database creation using DBCA to fail with the ORA-28040: No matching authentication protocol error in 12.1 Oracle ASM and Oracle Grid Infrastructure environments.

Workaround:  Set SQLNET.ALLOWED_LOGON_VERSION=8 in the $crs_home/network/admin/sqlnet.ora file.

Use the workaround before running 10.2.0.5 DBCA to create a database using 12.1 Oracle ASM and Oracle Grid Infrastructure.

3.24.15.11 Bug 14148942

If you are using Oracle Grid Infrastructure and you want to create an Oracle RAC release 11.1.0.7 database, then you may need to increase the DBCA default for session processes. For Oracle Database 12c, DBCA sets the default value for processes to 300. In earlier releases, DBCA set the default value to 150.

Workaround: If you see the error message ORA-00018:maximum number of session exceeded, then change the default value for session processes in DBCA to 300. DBCA will then successfully create the release 11.1.0.7 database to use with Oracle Grid Infrastructure release 12.1.

3.24.15.12 Bug 13479980

For installer invocation, DBCA in silent mode will display the following message and execution will stop after a validation warning. The default DBCA behavior is to stop after the following warning:

There are not enough servers available to allocate to this server pool, 
Database instances may not come up on specified cardinality. Do you want to continue? 

If you click Yes, the DBCA fails.

Workaround:  Before starting the installer, ensure that there are a sufficient number of servers in the free server pool. The number of free servers should be more than or equal to the cardinality specified in the installer for configuring the policy-managed Real Application Clusters database. The status and membership details of server pools can be checked using the following command:

Grid_home/bin/crsctl status serverpool

3.24.15.13 Bug 8288940

Creating pre-11.2 Oracle RAC database in 12.1 Oracle Grid Infrastructure for a cluster environment using DBCA may fail with following messages.

When using a cluster file system as storage, you see the following message:

ORA-00119: invalid specification for system parameter REMOTE_LISTENER

When using Oracle ASM as storage, you see the following message:

DBCA could not startup the ASM instance configured on this node

Workaround: Apply the patch for this bug in pre-11.2 database home. This patch is needed for 10.2.0.4, 11.1.0.6 and 11.1.0.7 database releases. No patch is needed for release 10.2.0.5.

3.24.16.1 Bug 15904575

When connected to pluggable databases, online translation may give unexpected errors or warnings if the application contains calls to Oracle-specific packages and procedures.

Workaround: Perform the translation by connecting to the root database of the multitenant container database (CDB).

3.24.17.1 Bug 16532807

The ORA-4036 error message is a known issue that affects customers in the very specific use-case scenario of creating a global CONTEXT index in parallel with a local CONTEXT index. Dropping and re-creating these indexes multiple times as well as enabling the BIG_IO option are known to manifest the problem. The symptoms of the problem include increased use of handle leaks.

Workaround: None.

3.24.17.2 Bug 15904575

The BIG_IO option of a CONTEXT index uses SecureFiles for the TOKEN_INFO column. When a CONTEXT index is created with the BIG_IO option, there is no check in place to make sure that the tablespace is Automatic Segment Space Management (ASSM). The TOKEN_INFO column of the $I table in the index is silently created with BasicFile.

Workaround: Ensure that the tablespace is set as ASSM before trying out the BIG_IO option.

3.24.17.3 Bug 14747404

With the CURSOR_SHARING parameter set to FORCE, queries using the ctxfiltercache operator involving save_score and topN will throw error DRG-10886.

Workaround: None.

3.24.17.4 Bug 14743270

When the BIG_IO option is specified and the AUTO_LEXER type with the MIXED_CASE attribute is set to YES, the following error might be returned during index creation:

ORA-00060: deadlock detected while waiting for resource

Workaround: Either do not use the MIXED_CASE attribute set for AUTO_LEXER or, if it is required, BIG_IO should not be specified.

3.24.17.5 Bug 14464880

Creating an index on a CTXSYS.CONTEXT index using the preference BIG_IO does not scale linearly with data size.

Workaround: Do not use the BIG_IO option.

3.24.17.6 Bug 14396559

The LONG VARCHAR2 (VARCHAR2 > 4K) feature is not supported with the CTXSYS.CONTEXT index.

Workaround: None.

3.24.17.7 Bug 14381004

Using the STAGE_ITAB and FORWARD_INDEX attributes on the same CTXSYS.CONTEXT index may cause internal errors.

Workaround: Try creating the index again.

3.24.17.8 Bug 14371181

DBMS_REDEFINITION.COPY_TABLE returns an internal error if used on a table which has a CTXSYS.CONTEXT index with BIG_IO and SEPARATE_OFFSETS attributes set.

Workaround: None.

3.24.17.9 Bug 13990319

In this release, the LONG VARCHAR2 (VARCHAR2 > 4K) feature is not supported with the CTXSYS.CTXCAT index.

Workaround: None.

3.24.18.1 Bug 17399376

Users are not stopped from installing Oracle Database Client 12c Release 1 (12.1.0.1.0) or Oracle Database Examples 12c Release 1 (12.1.0.1.0) software, using the respective installers, into an existing Oracle home that is of a different version.

Workaround:  Before installing release 12.1.0.1.0 Client and release 12.1.0.1.0 Examples software into an existing Oracle home, verify that the version of the Oracle Database server is also release 12.1.0.1.0.

3.24.18.2 Bug 16217171

When installing Oracle Grid Infrastructure, the Oracle Universal Installer (OUI) may hang and following error can be seen in log file:

Exception in thread "AWT-EventQueue-0" java.lang.OutOfMemoryError: PermGen space 

Workaround:  Rerun the installation.

3.24.18.3 Bug 16200062

Oracle Grid Infrastructure 12c Release 1 (12.1) eliminates the need to allocate local storage space on each node for the Cluster Health Monitor (CHM) repository. In 12.1, this repository, now known as the Grid Infrastructure Management Repository (GIMR), resides in the same disk group as the Oracle Cluster Registry (OCR) and voting disk files. The required space for the GIMR depends upon the current and maximum size of the cluster as well as on the level of redundancy for this disk group. The required space available needs to be manually checked as explained in the "Requirements for Oracle Grid Infrastructure Shared File System Volume Sizes" section of the Oracle Grid Infrastructure Installation Guide. This space availability is not checked automatically as part of the Oracle Universal Installer (OUI) upgrade installation. If insufficient space is found, the upgrade will fail when trying to create the GIMR.

Workaround:  The solution is to add additional disks to the disk group and retry the OUI step that creates the GIMR.

3.24.18.4 Bug 16177073

Mixed case host names are not supported by the Oracle Universal Installer (OUI) for Oracle RAC or Cluster Ready Services (CRS) homes.

Workaround:  Do not use mixed case host names.

3.24.18.5 Bug 16076425

The pound sign character (#) should not be specified in the PDB name field in the installer.

Workaround:  None.

3.24.18.6 Bug 15957211

Deinstalling the database may fail with the following error message if Oracle Grid Infrastructure and the database are installed with different users:

./runInstaller[192]: /tmp/deinstall_bootstrap/deinstallbootstrap.log: 
 cannot create [Permission denied]

Workaround: Manually remove deinstall_bootstrap and rerun the deinstallation to remove the Grid home.

3.24.18.7 Bug 15907640

During upgrade, when ignore unreachable nodes is selected in the OUI and there are some nodes down, a force upgrade cannot be done if the nodes that were down come back up before the force upgrade command (for example, rootupgrade.sh -force) completes.

Workaround: Make sure the nodes that were down remain down until the upgrade completes.

3.24.18.8 Bug 13874267

The Oracle Universal Installer (OUI) may fail to install Oracle Grid Infrastructure due to the allocation of too many shared memory segments. This problem is likely to occur only on the first start of the OUI after a server is started.

Workaround: Before finalizing the Oracle Grid Infrastructure installation (before clicking the Install button on the last page of the OUI dialog), issue an ipcs -m | wc -l command on another terminal to check the number of allocated shared memory segments at this moment in time.

If the returned number of allocated shared memory segments significantly exceeds 500, cancel the current OUI installation and restart runInstaller, because the installation is likely to fail. You can save the response file in order to resume the installation using a response file. Otherwise, you will have to reenter all of the required information during the dialog again. Since the problem is likely to occur only on the first start of the OUI after a server was started, restarting the OUI without restarting the server should solve the problem.

3.24.18.9 Bug 12930328

If the central inventory location is different on different nodes of a cluster, addnode.sh does not update the inventory correctly on remote nodes of the cluster.

Workaround: Adding nodes to a cluster requires the central inventory location to be the same on all the nodes of the cluster. Please ensure that this is the case prior to running addnode.sh.

3.24.18.10 Bug 12885716

If you install a 32-bit Oracle database and a 64-bit Oracle database in the same ORACLE_BASE, it may lead to unexpected results when using the deinstallation tool to remove one of the databases. The deinstallation tool removes all of the Oracle homes under the ORACLE_BASE if these Oracle homes do not use the same central inventory.

Workaround: Avoid using multiple central inventories. Do not use the same ORACLE_BASE for 32-bit and 64-bit database installations or always perform the deinstallation from a 64 bit home.

3.24.18.11 Bug 8666656

The Oracle Universal Installer (OUI) runInstaller script that resides in the Oracle home (ORACLE_HOME/oui/bin/runInstaller) cannot be used to install the Oracle Database 12c releases, Oracle Grid Infrastructure for a cluster, and Oracle Database Client.

Workaround: Use Oracle Universal Installer on the respective Oracle Database 12c product media to install each product.

3.24.19.1 Bug 16069266

Using Transportable Tablespaces (TTS) to export or import tables with Binary XML data is not supported.

Workaround: Use the Oracle Data Pump conventional path to move data.

3.24.19.2 Bug 16055386

If you plug an Oracle Database 12c non-CDB database, which has an XMLType table with an XMLIndex defined, into an Oracle Database 12c CDB database and run the noncdb_to_pdb.sql script, the XMLIndex could become disabled.

Workaround: Manually enable the XMLIndex by issuing the following DDL:

ALTER INDEX <index_name> ENABLE;

3.24.19.3 Bug 15858259

Consider the following scenario:

On an 11.2.0.2 database, you create an XMLType table or a table with an XMLType column that is partitioned and then create a structured XMLIndex on the table. Next, you upgrade the 11.2.0.2 database to Oracle Database 12c or above. If you execute the ALTER TABLE EXCHANGE PARTITION statement on one of the base table partitions with another partition or table that is created in the Oracle Database 12c database with the same structure and similar XMLIndex, then it will fail with the following error:

ORA-64123: XMLIndex DDL: failure of a recursive DDL 
ORA-14097: column type or size mismatch in ALTER TABLE EXCHANGE PARTITION 

This is not an issue if your index was created on a non-11.2.0.2 database. 11.2.0.2 is the only version with this issue. Other versions like 11.2.0.1 and 11.2.0.3 do not have this issue.

Workaround: After upgrading to Oracle Database 12c or above, drop and re-create the structured XMLIndex on the table where the index was upgraded from 11.2.0.2.

3.24.19.4 Bug 14675056

When 32K VARCHAR data type is enabled and XML storage is object-relational, UPDATEXML or equivalent XQuery Update operation may run into a parser issue.

Workaround: None.

3.24.19.5 Bug 14675052

When 32K VARCHAR data type is enabled and XML storage is object-relational, selecting the XMLType column directly may return incorrect data. For example:

SELECT xml_col FROM some_table p;

Workaround: Call the GETCLOBVAL() function. For example:

SELECT p.xml_col.getclobval() FROM some_table p;

3.24.19.6 Bug 14578945

Prior to Oracle Database 12c Release 1 (12.1), when inserting an XML element without a name space into an XML document, the newly inserted element is assigned to the name space of the parent element if the default name space is defined in the parent element, which is not correct.

In Oracle Database 12c, the new element will be inserted with xmlns="".

Workaround: None.

3.24.19.7 Bug 14243273

An XML Query containing CLOB, BLOB or NCLOB data types as bind variables is not supported for logical standby, Oracle GoldenGate and XStream replication.

Workaround: Run the query using data types other than CLOB, BLOB or NCLOB.

3.24.19.8 Bug 12991540

For logical standby based schema registration replication, compatibility should be set to Oracle Database 12c or above. For XStream and Oracle GoldenGate, the schema has to be individually registered on each instance and replication is not supported. All of the procedures in DBMS_XMLSCHEMA package are supported except COPYEVOLVE and COMPILESCHEMA.

Workaround: None.

3.24.19.9 Bug 12868171

Supplemental logging is unsupported for an XMLQuery update with variables bound to a REF cursor.

Workaround: Before updating XMLType columns or attributes that need replicating, store the evaluation of REF cursors in non-cursor variables, and then update the columns or attributes with these variables instead of the REF cursors.

3.24.19.10 Bug 9648042

The EXCHANGE PARTITION operation does not work for XMLType tables that have an XMLIndex structured component.

Workaround: To workaround this problem, take the following steps:

1. Before executing the EXCHANGE PARTITION operation, issue the following call:

   EXEC DBMS_XMLSTORAGE_MANAGE.EXCHANGEPREPROC(user, 'partitioned_myxml');
   

   partitioned_myxml is your partitioned XMLType table.

2. Execute the EXCHANGE PARTITION operation. For example:

   ALTER TABLE partitioned_myxml EXCHANGE PARTITION p1 WITH TABLE myxml;
   

3. After executing the EXCHANGE PARTITION operation, call the following procedure:

   EXEC DBMS_XMLSTORAGE_MANAGE.EXCHANGEPOSTPROC(user,'partitioned_myxml');
   

If the myxml table also has a structured XMLIndex on it, you also need to execute the DBMS_XMLSTORAGE_MANAGE.EXCHANGEPREPROC and DBMS_XMLSTORAGE_MANAGE.EXCHANGEPOSTPROC procedures on myxml table. For example:

1. Before executing the EXCHANGE PARTITION operation, execute the following:

   EXEC DBMS_XMLSTORAGE_MANAGE.EXCHANGEPREPROC(user, 'myxml');
   

2. After executing the EXCHANGE PARTITION operation, execute the following:

   DBMS_XMLSTORAGE_MANAGE.EXCHANGEPOSTPROC(user,'myxml');
   

3.24.19.11 Bug 8687128

An Oracle RAC system allows multiple concurrent database instances to share a single physical database. However, dispatches for Oracle XML DB in an Oracle RAC database do not listen on the virtual IPs.

Workaround: To enable Oracle XML DB to use TCP(S) on an Oracle RAC system, you must configure the TCP(S) dispatchers for each database instance of the cluster as follows (where SID is the SID of the instance and HOST is the host name for the physical database):

SID.dispatchers="(address=(protocol=tcps)(host=HOST-vip)(service=SIDxdb))" 

For non-secure dispatchers (TCP, not TCPS), use tcp in the command instead of tcps.

3.24.20.1 Bug 16903140

A new feature was added in Oracle Database 12c Release 1 (12.1) to support change tracking across resetlogs during the execution of an ALTER DATABASE OPEN RESETLOGS statement. This is achieved by deleting bitmaps that are not required because they cover changes that were made after the specified resetlogs point-in-time. There is an issue with the bitmap deleting that can cause error ORA-600 [krccchs_1] during OPEN RESETLOGS.

Workaround: Disable and reenable change tracking. You must disable change tracking before executing the ALTER DATABASE OPEN RESETLOGS statement, and then re-enable change tracking after the open resetlogs operation is complete.

3.24.20.2 Bug 16187090

In order to backup a database in NOARCHIVELOG mode, the database must be cleanly shutdown and then started in a mounted state.

When a multitenant container database (CDB) is in NOARCHIVELOG mode and the root is open, an RMAN backup will fail (as expected), but will still generate backup pieces. These backup pieces are not usable for CDB or PDB recovery because they were generated while the CDB was in NOARCHIVELOG mode and open.

Workaround: Use one of the following as a workaround to this situation:

• Ensure CDB is cleanly shutdown and then started in a mounted state before taking a backup.

• Ensure CDB is in ARCHIVELOG mode before taking backup.

Note that the unusable backups originally created should be manually deleted using RMAN.

3.24.21.1 Bug 8256753

A connect using SCAN and EZCONNECT on one client machine can be requested to use a specific SCAN listener. Therefore, load balancing by round-robin DNS is not possible.

Workaround: Connect to a database using the following configuration specifying LOAD_BALANCE=on in tnsnames.ora:

ORCL = 
   (DESCRIPTION = 
     (LOAD_BALANCE=on) 
     (ADDRESS = (PROTOCOL = TCP)(HOST = stscan1)(PORT = 1521)) 
     (CONNECT_DATA = 
       (SERVER = DEDICATED) 
       (SERVICE_NAME = srv.world) 
     ) 
   )