My experience with the latest SAP Bundle Patches wasn’t very good. In a previous post I mentioned this problem: http://sapnotebook.com/lob-segment-growing-large-bug-sap-bundle-patch-12-1-0-2-170418-201705/.
And now there is another bug that I experienced: 24447296 – ORA-3171 in a Non-ADG Database (Doc ID 24447296.8).
This might also be happening to you if you installed one of these patches: SBP 201711, 201708 or 201802 for Oracle 12.1.0.2 and you have a UNIX environment.
The bug manifests right after installing one of these patches by filling the system with short dumps of the type ‘DBIF_RTAB_SQL_ERROR’ with short text SQL error 3171 when accessing table “VBDATA”.
And when analyzing the developer traces for the corresponding work processes, you will get something similar to this:
———————————————-
C Wed Apr 11 08:43:46 2018
C check2: OCIStmtExecute() failed with -1=OCI_ERROR
C SQL error 3171:
C ORA-03171: Recovery was terminated during query execution
C *** ERROR => Error 3171 in stmt_execute() from oci_execute_stmt(), orpc=0
[dbsloci.c 17449]
C {root-id=0050568F7B7D1ED88FAA2DC3B5D6D34C}_{conn-id=00000000000000000000000000000000}_0
C *** ERROR => ORA-03171 occurred at SQL stmt (parse error offset=0)
[dbsloci.c 17450]
C {root-id=0050568F7B7D1ED88FAA2DC3B5D6D34C}_{conn-id=00000000000000000000000000000000}_0
C Dump statement cache (after SQL error):
———————————————-
VBDATA is not the only table for which the dumps can appear, also the tables TST03, SMW3_BDOC2, BC_MSG, SXMSCLUR, IUUC_RL_DATA, SXMXCLUP, SOFFCONT1, /WATP/TMOBMRIA can be affected.
The error does not occur in all the systems that have these patches installed, this is most likely to happen in the systems where the above tables are very large and get accessed very often.
The tricky part is that the bug can manifest directly into production system, even if the patch was previously applied to DEV and QAS and no side effects appeared, because in production the load on the tables can be significantly higher.
After some time the error has been also documented by SAP in two notes, which I recommend to be read very careful and also to be checked from time to time for updates:
2633327 – ORA-03171 while accessing table in a Non Active Data Guard database (non-ADG)
2668860 – ORA-00600 [3020] with block type NGLOB: “Hash Bucket”
In the first versions of the note 2633327, it was stated that the bug would get fixed by applying some Oracle merge patch or by installing a newer bundle patch 201805.
But a newer update of the note states that you should definitely not install those mentioned patches, as you could get into even more serious problems, that can lead to an unrecoverable database!
At the moment there is not fix available and an update of the note is still waited to come.
So I would say think twice before installing any new bundle patch from Oracle as these can have a negative impact on your business jobs and can seriously affect the daily operation.
*** Update: On 15.08.2018 SAP updated the note 2633327 and several fixes are available depending your SBP installed. Read it carefully an apply the fix that matches your case.