Re: Lengthening backup
- From: Fuad Arshad <fuadar@xxxxxxxxx>
- To: terrysutton@xxxxxxx, oracle-l@xxxxxxxxxxxxx
- Date: Sat, 8 Jul 2006 17:41:59 -0700 (PDT)
terry,
try to do debug from rman.
rman provides a fairly lengthy debug trace file of the pl/sql it executes. you
might be hitting some bug in the 8.1.7 code execution
of dbms_backup_restore
----- Original Message ----
From: Terry Sutton <terrysutton@xxxxxxx>
To: oracle-l@xxxxxxxxxxxxx
Sent: Saturday, July 8, 2006 6:03:35 PM
Subject: Re: Lengthening backup
I did a short test, backing up 2 datafiles, and tracing the session on the
target DB. I found that, when the backup took about 30 minutes, over 1700
seconds were spent waiting on "SQL*Net message from client" during the
execution of
begin sys . dbms_application_info . set_session_longops ( rindex => :rindex
,
slno => :slno , op_name => :op_name , context => :context , sofar =>
:sofar
, totalwork => :totalwork , units => :units ) ; end ;
So this seems to be the culprit. Anyone have any idea why running
dbms_application_info.set_session_longops would 1) take so long and 2)
apparently take longer as the DB is up longer?
BTW, I also flushed the shared pool before one of the tests, and it had no
major effect.
--Terry
Andy,
The archivelogs are not being backed up by RMAN. And I'm seeing nothing in
v$session_longops:
SQL> select * from v$session_longops where sid in (918,1185);
no rows selected
One thing I did notice, FWIW, is at 10:22:35 yesterday, I looked in the
directory to which the backup pieces are written, and no files had been touched
since 10:19. Later, at 10:32:22, the latest file timestamp was 10:28. So it
seems like the backup is doing something between file writes. Don't know if
that helps any.
--Terry
Hi Terry, two things come to mind. How are the archivelog files being handled
(i.e. are they being bundled into the backup, perhaps without a delete?) and
does v$session_longops reveal any clues when the backup is running?
Andy Rivenes
arivenes@xxxxxxxx
At 11:05 AM 7/7/2006, Terry Sutton wrote:
I'm having some issues with a client's RMAN backup. They're on Oracle 8.1.7.4,
Solaris 8. It's a weekly full backup, and they're backing up ~1.5TB to disk,
and the longer the instance is up, the longer the backup takes. A recent
progression has been:
Other related posts:
