Re: [Fwd: Re: 10046 trace - unaccounted for time]

The unaccounted for time appears *after* the INSERT finished inserting one row.

EXEC #11:c=0,e=10076,p=1,cr=1,cu=7,mis=0,r=1,dep=0,og=4,tim=5188376817822

Do you have access to the code? Can you check what code follows the INSERT statement. The unacc time does not have to be caused by a cursor. For example, it could be running plsql code, and that would not show up in the sql trace file.

Rakesh

Managing Principal
dbPerf Inc.

On Wed, Aug 6, 2008 at 2:27 PM, Keith Moore <kmoore_at_zephyrus.com> wrote:
> Ok, we got our test system setup and turned on system wide tracing and now
> have the SQL statement where the hang occurs. The statement is querying
> all_constraints but it is not recursive SQL. It is being called from the
> application.
>
> Below is a relevant section of the trace file. It looks like it reuses CURSOR
> #12 but the 2nd query is on cdef$ which is also related to constraints. Not
> sure I understand that.
>
> It's also still not clear to me whether the unaccounted for time is for Cursor
> #12 or Cursor #11 or from something outside this session.
>
> This problem has now been reproduced on a 2nd server with a duplicate of the
> database. This server is very lightly loaded so I don't think it is server
> related.
>
> Keith
>

--
http://www.freelists.org/webpage/oracle-l