2 minutes to parse "select 1 from dual" sounds like a memory problem - it
may be that the problem will disappear if you restart the instance.
Before doing that, though, I'd enable extended (level 12, i.e. waits and
binds) sql-trace and repeat the explain plan for select 1 from dual to see
if the resulting trace file showed me exactly what happened just before the
error occurred - I'd guess that some recursive SQL that was trying to load
something into the dictionaty cache was running into a problem. If that
trace doesn't give you any clues I'd try the same with the simplest
possible statement that does cause the ORA-00600.
Regards
Jonathan Lewis
On Wed, 10 Mar 2021 at 11:00, Pap <oracle.developer35@xxxxxxxxx> wrote:
Hi All, Oracle version is - 12.1.0.2.0 and its Exadata. I am seeing
Ora-600 while fetching explain plan through "set autotrace traceonly
explain". Even explain plan for "Select 1 from dual" is working but that is
also taking ~2minutes and that too failing with the same Ora-600 for other
queries. I got a related bug from mos, but the work around which is
provided as setting "_optimizer_push_pred_cost_based" to false is also
not working, which means we are hitting some other bug. Has anybody
encountered such a scenario?
Explain Plan for Query Errors with 'ORA-00600: internal error code,
arguments: [KGL-heap-size-exceeded]' (Doc ID 2469727.1)
SQL> select 1 from dual;
Execution Plan
----------------------------------------------------------
ERROR:
ORA-00600: internal error code, arguments: [ktrenvchk_1], [0], [1], [0],
[1],
[0], [0], [0], [], [], [], []
SP2-0612: Error generating AUTOTRACE EXPLAIN report
