No host can be created on SM55

Question:

Couldn't create any host via SM55. However, this only happen in SAP Solution Manager, in other SAP systems, it is working fine.

Error says "No entries in the table THOST available".

Solution:

Error in SM55, could caused by an empty table.

Please create some dummy entries in the database table THOST via transaction SE16, then SM55 should run as usual.

For additional information, here are some SAP OSS Notes to refer to:
#10743 - Host name longer than 8 characters
#25941 - System does not find host name

How to cancel online backup

Question:

How to cancel online backup which was scheduled via DB13 for SAP ECC6 and database on Oracle 10g?

Solution:

Tthe only way to cancel a backup is to kill the session. Please refer to SAP OSS Note #4162 regarding consequences of killing a backup session while running.

Backup after upgraded ECC 6.0.0 to 6.0.4

Question:
Upgrade on ECC 6.0 0. After successfully upgrade, but unfortunately backup is not possible. Backup was still possible before upgrade. Error on backup log from DB13.
tape. hdr0, I/O-Error.

Solution:
If you are experienced error and cannot create <.tape.hdr0> (errno:13).tape.hdr0 10 blocks
Please check and make sure the permissions of BR*Tools have been setup correctly as per SAP OSS Note #113747.

Missing Tables and Indexes in DB02

Question:

In DB02 transaction diagnosis -> Missing Tables and Indexes has 367 missing optional indexes.

How do I solve this?

Solution:

For optional indexes that have not been created, these optional indexes are delivered by SAP. It is
recommend to ignore the optional indexes because they do not exist on database level.

Now you may ask, what is the use of optional index and where are they required? If you have performance problems while accessing the tables and once you have identified the table, you can check whether these
optional indexes delivered by SAP will enhance the performance by creating them and testing them in QAS or DEV system. If they help to enhance the performance, you can create them in production. If you are
not experiencing any issues, there is no need to create these indexes. Sometimes, these indexes will be created based on the database being used and they are only applied in some specialized functions/situations.
FYI: Opional indexes are stored in the SDBMONDDDB table.

This is an example of how it works:

- Transaction SE11.
- Enter the table(for example) ADRC, choose index of G.  You will see the
   index is marked as 'Non-unique Index' and 'No database index'.
Click on the 'Database system' icon next to it, you will see it is
applicable to the selected database shown there.

If you experience performance problems with these missing indexes,
you may create them by following:

- Transaction DB02
- Checks
- Database <-> ABAP Dictionary
- Recheck
- Click over the index you wish create.
- Click on "Display definition Button".
- Click on "Display & Change" button.
- If the system request a change request or ask for maintain in
  original language, please provide this info.
- Deselect the radio button 'No database index' and select 'Index on
  all database systems' or 'For selected database systems'.
- Save and activate. The index will be created. To refresh the
  missing index list, please repeat the steps 1 - 4.

Core Dump on ORA errors ORA-00604 ORA-04031

Frecuently, generating a lot of dumps with this ORA errors and refreshed the shared memory, due to insufficient memory for allocation.

ORA-00604: error occurred at recursive
SQL level 2#

ORA-04031: unable to allocate 4120 bytes of shared memory
("shared pool","select/*+rule*/ bucket_cn...","Typecheck","seg:kggfaAllocSeg")"
[RSQL/FTCH/SNAP_BEG ]

Attempt to increased all SGA memories at the machine's physical limits, and the error occurs again.

Solution:

There is an error cause by ORA-4031.
Please refer to SAP Note #1120481: ORA-4031 - memory allocation errors
for object-level.

Reason and Prerequisites
This is the Oracle bug 6487875/5573238, which is corrected with Oracle
10.2.0.4. This problem may occur on databases that have a large number of objects (for example, a lot of partitions).

Up to Oracle 10.2.0.4, if the above problem occurs, please set the
following init<SID>.ora parameter as a workaround
      _object_statistics=false

Also, set the parameter online, as follows:
   alter system set "_object_statistics"=false scope=spfile;

Then always restart the database to activate the parameter.

Here are some additional information regarding this on SAP OSS Note,
#839187: Oracle 10.2.0: Applying patch set/patches/patch
#23345: Consistency check of ORACLE database

HP-UX Oracle Alert Log ST04 ORA-07445: Exception encountered: Core dump [$cold_qerfxArrayMaxSize ()+14320]

Question:

Operating System HP-UX v11.3, in Oracle Alert Log (Transaction ST04) is specified into message the

following irregular distances:

Errors in file

ORA-07445: Exception encountered: Core dump [$cold_qerfxArrayMaxSize ()+14320] [Address not mapped to object] [0x0]

By which the error is caused is not to be reproduced.

Solution:

Regarding the error ORA-07445: [$cold_qerfxArrayMaxSize()+14320], please

 refer to SAP Note #972574 for solution.

For Operating System HP-UX v11.3, please refer to SAP OSS Note #1031960 for

Recommendations and availability of HP-UX 11.31, and SAP OSS Note #837670

for Minimum OS patch recommendations for HP-UX.

According to Oracle Metalink, error ORA-07445 is a problem which is

reported in bug:5382798 and found to be an issue with the C compiler

used on HP Itanium to compile the Oracle 10.2 code. Oracle Development

has released a patch:5442780 that contains re-compiled version of the

affected module (compiled without the feature that is causing the

problem).

Please proceed to Oracle Metalink and download said patch and install.

Not able to archive the Oracle logfiles with RMAN and Legato Networker for Oracle Release 10.2.0.2

Question:

Not able to archive the Oracle logfiles with RMAN and Legato Networker for Oracle Release 10.2.0.2. Could it be a problem with the xfs-Filesystem?

SuSE/Novell: Linux, Filesystem & I/O Type Supportability [ID 414673.1]

WARNING: Using unsupported filesystems and/or options for Oracle Database
or using XFS due to an undocumented defect can cause redo log corruption
which can only be detected after it happens when doing a restore or dump of redo log to trace file.

ORA-00283: recovery session canceled due to errors
ORA-00368: checksum error in redo log block
ORA-00353: log corruption near block 4096 change 21180175 time 03/10/2008 00:14:53

Solution:

After analyzed the system log, it is recommended to refer to SAP OSS Note
#335562: ORA-7445 collective note regarding the solution of this matter.

Also, refer to SAP OSS Note #23345 for Consistency check of ORACLE database.

Running several Oracle 10.2.0.2 Instances on a single Linux host

Question to SAP Solution:

What happen when running several Oracle 10.2.0.2 Instances on a single Linux host but need to patch only one of them.

What happen with the common Oracle files such as in /oracle/client, /oracle/oraInventory and etc?

Similar like in SAP OSS Note #964783.

Solution:

Regarding inquiry on SAP OSS Note #964783, it is clearly stated that it
applies only for Oracle release 9.2 on UNIX/Linux. Therefore, it is not
applicable for Oracle 10.2.0.2 as having in the system.

However, there is only one instant client provided by SAP which is
version 10.2.0.4 therefore what the note proposes is not possible with
the instant client.

Please refer SAP OSS Note #19829 - Oracle Instant Client
 Installation and Configuration.

Having multiple SAP systems with the same kernel version then
able to install one instant client and have all instances pointing to
the same client path.