Subject: Batch load freezes every couple of minutes

Content-Type: text/plain;

        charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable Subject: Batch load freezes every couple of minutes=20

RDBMS Version: 9.2.0.5 RAC
Operating System and Version: Sun Solaris 5.8 Error Number (if applicable):=20
Product (i.e. SQL*Loader, Import, etc.):=20 Product Version:=20

Batch load freezes every couple of minutes

Instance runs 20 parallel load processes. Load completely stops every=20 couple of minutes ( there are no 'user commits' for up to 2 minutes in = v$sysstat ).=20
This is output (edited to show TX lock only). Instance is up just couple = of days.=20

1 select eq_type "Lock",=20
2 total_req# "Gets",=20
3 total_wait# "Waits",=20
4 cum_wait_time=20
5 from v$enqueue_stat=20
6* where Total_wait# > 0=20

Lo Gets Waits CUM_WAIT_TIME=20
TX 30779423 30775582 67592703=20

Figure for waits looks huge to me.=20

My tests show that Waits ( total_wait# ) only gets increased when there = is row level lock contention ( session is waiting for commit/rollback = from the other session ). Average wait time calculated from = v$system_event for 'enqueue' is very small.=20 Could this high figure for total_wait# be the reason why load freezes ?=20

( Checkpoint not complete are the problem, but I am now looking beyond = that ).

Regards, Ranko.=20

    =20

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