DBIF_SETG_SQL_ERROR after upgrade to Release 10.2.0.2.0 Solaris 5.10

[Former Member]

I get short dumps in our DEV ERP 2004 / 640 sap kernel patch level 196 system occasionally after upgrading from 9i to 10g. Short dumps state u201CDBIF_SETG_SQL_ERROR ORA-03106: fatal two-task communicationu201D, and this usually involves having to go into SM13 and repeating updates. I have an ongoing open call for quite a long time now with no resolution.

I read this thread

I used mopatch to apply all the recommended patches in note 817096. Also, after reading note 1005758, I also including tried including patch 5092134. According to the Mopatch log, it stated that note 5092134 is a superset of patch 5099986. Hereu2019s a list of current patches on our DEV system:

Patch 6729801 : applied on Thu Jun 12 18:59:42 EDT 2008

Patch 6455161 : applied on Thu Jun 12 18:56:53 EDT 2008

Patch 6435823 : applied on Thu Jun 12 18:56:11 EDT 2008

Patch 5941030 : applied on Thu Jun 12 18:52:39 EDT 2008

Patch 5638146 : applied on Thu Jun 12 18:20:54 EDT 2008

Patch 5636728 : applied on Thu Jun 12 18:08:45 EDT 2008

Patch 5618049 : applied on Thu Jun 12 18:08:01 EDT 2008

Patch 5571916 : applied on Thu Jun 12 18:07:20 EDT 2008

Patch 5530958 : applied on Thu Jun 12 18:06:39 EDT 2008

Patch 5442919 : applied on Thu Jun 12 18:05:57 EDT 2008

Patch 5369855 : applied on Thu Jun 12 18:05:15 EDT 2008

Patch 5363584 : applied on Thu Jun 12 18:04:34 EDT 2008

Patch 5345999 : applied on Thu Jun 12 18:03:53 EDT 2008

Patch 5103126 : applied on Thu Jun 12 18:03:12 EDT 2008

Patch 5063279 : applied on Thu Jun 12 18:02:31 EDT 2008

Patch 4952782 : applied on Thu Jun 12 18:01:50 EDT 2008

Patch 4899479 : applied on Thu Jun 12 18:01:08 EDT 2008

Patch 4883635 : applied on Thu Jun 12 18:00:20 EDT 2008

Patch 4742607 : applied on Thu Jun 12 17:59:35 EDT 2008

Patch 4668719 : applied on Thu Jun 12 17:58:54 EDT 2008

Patch 5090361 : applied on Thu Jun 12 17:58:07 EDT 2008

Patch 4518443 : applied on Thu Jun 12 17:57:26 EDT 2008

Patch 3748430 : applied on Thu Jun 12 17:57:01 EDT 2008

Patch 5117016 : applied on Thu Jun 12 17:56:14 EDT 2008

Patch 5099986 : applied on Mon Jun 02 19:05:54 EDT 2008

Patch 7021805 : applied on Tue May 27 18:08:23 EDT 2008

Patch 6340979 : applied on Wed May 07 21:15:12 EDT 2008

Patch 6819785 : applied on Wed May 07 20:47:25 EDT 2008

This problem doesnu2019t occur on our Test/Qa system, which doesnu2019t have as many patches. This is a list of our Test patches (also upgraded to 10G):

Patch 6394997 : applied on Sun Jun 29 08:47:18 EDT 2008

Patch 6435823 : applied on Mon Jun 23 20:05:13 EDT 2008

Patch 6340979 : applied on Mon Jun 23 19:23:41 EDT 2008

Patch 5941030 : applied on Mon Jun 23 19:18:13 EDT 2008

Patch 5636728 : applied on Mon Jun 23 19:16:53 EDT 2008

Patch 5618049 : applied on Mon Jun 23 19:16:13 EDT 2008

Patch 5571916 : applied on Mon Jun 23 19:15:33 EDT 2008

Patch 5530958 : applied on Mon Jun 23 19:14:53 EDT 2008

Patch 5363584 : applied on Mon Jun 23 19:12:54 EDT 2008

Patch 5103126 : applied on Mon Jun 23 19:11:35 EDT 2008

Patch 5063279 : applied on Mon Jun 23 19:10:56 EDT 2008

Patch 4952782 : applied on Mon Jun 23 19:10:16 EDT 2008

Patch 4668719 : applied on Mon Jun 23 19:07:24 EDT 2008

Patch 4518443 : applied on Mon Jun 23 19:06:02 EDT 2008

Patch 3748430 : applied on Mon Jun 23 19:05:38 EDT 2008

Patch 5117016 : applied on Mon Jun 23 19:04:58 EDT 2008

My LD_LIBRARY_PATH is set to LD_LIBRARY_PATH=/oracle/DEV/102_64/lib:/usr/sap/DEV/SYS/exe/run

And ORA_NLS33=/oracle/client/92x_64/ocommon/nls/admin/data

Thanks in advance

[Former Member]

Upgraded Oracle to 10.2.0.4 and latest patches for Solaris as well as the 640_EX2 kernel

[vince_laurent]

Can you post the ENV and SET of the user? Sounds like you might have TWO_TASK set up somewhere.

Vince

[Former Member]

Hi Vince, here ya go:

sapdev:oradev 100% env

HOME=/oracle/DEV

PATH=/oracle/DEV/102_64/bin:/oracle/DEV:/usr/sap/DEV/SYS/exe/runU:/usr/sap/DEV/SYS/exe/run:/usr/bin:/usr/ccs/bin:/usr/ucb

LOGNAME=oradev

HZ=100

TERM=vt100

TZ=US/Eastern

SHELL=/usr/bin/csh

MAIL=/var/mail/oradev

PWD=/tmp

USER=oradev

SAPSYSTEMNAME=DEV

DIR_LIBRARY=/usr/sap/DEV/SYS/exe/run

LD_LIBRARY_PATH=/oracle/DEV/102_64/lib:/oracle/DEV/102_64/lib:/usr/sap/DEV/SYS/exe/run:/oracle/DEV/102_64/lib64

THREAD=NOPS

dbms_type=ORA

dbs_ora_tnsname=DEV

dbs_ora_schema=SAPDEV

ORACLE_SID=DEV

ORACLE_HOME=/oracle/DEV/102_64

ORACLE_BASE=/oracle

SAPSID=DEV

NLS_LANG=AMERICAN_AMERICA.WE8DEC

SAPDATA_HOME=/oracle/DEV

ORA_NLS10=/oracle/DEV/102_64/nls/data

EDITOR=vi

LOGINFROM=10.29.2.113

DISPLAY=10.29.2.113:0.0

sapdev:oradev 102% set

DBSID DEV

argv ()

autologout 0

cwd /tmp

d /oracle/DEV/102_64/lib64

filec

history 50

home /oracle/DEV

hostname sapdev

i 1

list ()

lscommand /bin/ls

mypath . /oracle/DEV/102_64/bin /oracle/DEV /usr/sap/DEV/SYS/exe/runU /usr/sap/DEV/SYS/exe/run /usr/bin

notify

p /oracle/DEV/102_64/lib64

path (/oracle/DEV/102_64/bin /oracle/DEV /usr/sap/DEV/SYS/exe/runU /usr/sap/DEV/SYS/exe/run /usr/bin /usr/ccs/bin /usr/ucb)

prompt sapdev:oradev !%

shell /bin/csh

status 0

term vt100

user oradev

sapdev:oradev 103%

[Former Member]

Oh one other thing, one thread mentioned increasing the shared_pool_size. It's currently set to 672M, but in our Test system it's set to 1568M. Could that be it?