Re: Innocent SQL causes an imminent process death..

Unemployed in the Valley wrote:

> I wonder if anyone ever encountered something like this:
>
> Environment - Oracle 8.1.6 on Solaris 2.8
> Table "relations" defines the node hierarchy.
>
> SQL> connect username/password_at_server
> Connected.
> SQL> desc relations;
> Name Null? Type
> ----------------------------------------- --------
> ----------------------------
> REC_ID NOT NULL NUMBER
> PARENT_NODE_ID NOT NULL NUMBER
> CHILD_NODE_ID NOT NULL NUMBER
> CREATION_DATE NOT NULL DATE
>
> SQL> select count(*)
> 2 from relations
> 3 start with parent_node_id=3
> 4 connect by prior child_node_id=parent_node_id
> 5 and prior child_node_id in (select 3 from dual)
> 6 ;
> select count(*)
> *
> ERROR at line 1:
> ORA-03113: end-of-file on communication channel
>
> -----------------
>
> I tried it on 3 different servers with the same results from Windows &
> Solaris clients.

95+% of the time ORA-03113 is Oracle's way of telling you that the SYSADMIN and DBA that set up the O/S and the Oracle database didn't RTFM.

Have one of them check the kernel values in /etc/system and compare them with the values required by Oracle which they can find in the readme file on the CD they appear to have not read and at http://docs.oracle.com.

But other possibilities exist ... one of which is that they haven't applied the Oracle patches to 8.1.6 that they should have but I can not tell from your posting. Straight 8.1.6.0.0 must be patched to be stable and perform anything close to what it should.

Daniel Morgan Received on Thu Oct 03 2002 - 14:47:36 CDT