PB 7.01 Oracle ORA-03113

p.blender

unread,

Oct 1, 1999, 3:00:00 AM10/1/99

to

Has anyone else experience Oracle error ORA-03113 when developing or running
an app under PB 7.01 (we have seen it with all builds).

ORA-03113 is a generic end of communications error with the server. It
happens mostly randomly and with a couple of cases we can reproduce.

The particulars are:

Oracle 7.3.4.4 Sun Solaris
SQLNET 2.3.4 both sides
Windows NT 4.0 SP4 and SP5
PB builds 5031, 6012, and 7015

We do have an open case with Sybase and Oracle. No help from either yet
(but Sybase was quick to blame Oracle).

Thanks!

Terry Dykstra

unread,

Oct 5, 1999, 3:00:00 AM10/5/99

to

No problems at all.
Does Oracle generate any trace files?
Potential culprits: invalid data, causing ora-600,causing ora3113; Listener
down; network traffic; corrupt table/indexes; complex functions can do it
too.

--
Terry Dykstra (TeamSybase)
Canadian Forest Oil Ltd.

p.blender wrote in message ...

p.blender

unread,

Oct 5, 1999, 3:00:00 AM10/5/99

to

Thanks for the reply. It have looked at the trace files on client and
server. They are not showing ORA-600, but do show some internal Oracle
errors. It's a nasty problem because we can reproduce it with the Oracle
driver or ODBC, but it's in the guts of OCI to Oracle.

We know it's not caused by database corruption or a dead listener. Network
traffic is ruled out because we have specific tests that can reproduce it.

Oracle originally called it a packet error, but they have now backed off
that since it can be reproduced. Sybase has blamed Oracle because they
claim it's an OCI level error.

My gut tells me that PB is feeding Oracle OCI bad information from time to
time. We have found cases where PB does not clean up the development
environment correctly after a program is stopped. That extra remaining
stuff hangs around and causes side-effect problems.

I will post any solutions we find. The scary thing is we are not getting
this with PB 6.5.

Are you currently working with Oracle 7.3.4.3 or Oracle 7.3.4.4?

Thanks

Terry Dykstra <tdyk...@cfol.ab.ca> wrote in message
news:g47y9a4D$GA...@forums.sybase.com...

Terry Dykstra

unread,

Oct 6, 1999, 3:00:00 AM10/6/99

to

I use 7.3.4.3.0. If it happens both with ODBC and PB native driver, it
looks like it is an Oracle OCI issue. But whay it doesn't happen with PB6.5
is then a mystery.

--
Terry Dykstra (TeamSybase)
Canadian Forest Oil Ltd.

p.blender wrote in message <3JV4#b5D$GA....@forums.sybase.com>...

Andreas Mehner

unread,

Oct 8, 1999, 3:00:00 AM10/8/99

to p.blender

Hallo,
we can provoke the ORA-03113 under PB 7.0.1-7015 under following
circumstances:

1.
CREATE OR REPLACE PROCEDURE select_emp(st VARCHAR2, rc IN OUT p_glob.rctyp)
AS
BEGIN
OPEN rc FOR
SELECT * FROM emp WHERE name=st;
END;
/

2.
create a datawindow based on this stored procedure
(the datawindow has an argument "ST" of type "string")

3.
recreate the stored procedure again with same or other code

4.
- retrieve the datawindow with retrieve("")
- this retrieve causes the ORA-03113

As a workaraound, we prevent that the argument for
the datawindow is the "empty string".
The error comes up only after recreating the stored procedure!

We don't know, if this is a PB or Oracle Bug!

Andreas Mehner
DELTA BARTH Systemhaus GmbH

p.blender

unread,

Oct 9, 1999, 3:00:00 AM10/9/99

to

Thanks for the example. I will try to do this in our setup and then see if
PB 6.5 does it.

Anything is ammo to take to Sybase.

We have done more research and now believe the problem is not Oracle. We
have situations were it appears that PB 7 is trashing something and causing
bad Oracle OCI calls.

Andreas Mehner <meh...@delta.de> wrote in message
news:37FDAA51...@delta.de...