On Oct 16, 9:14 am, csn..._at_gmail.com wrote:
> On Oct 16, 7:24 pm, Ben <bal..._at_comcast.net> wrote:
>
> > On Oct 16, 7:20 am, Ben <bal..._at_comcast.net> wrote:
>
> > > On Oct 16, 4:06 am, csn..._at_gmail.com wrote:
>
> > > > To the OP, yes it does sound like fet$/uet$ contention, and there is
> > > > nothing you can do except be patient and wait, in the knowledge that
> > > > you are getting rid of the problem, slowly but surely as you move to
> > > > LMT.
>
> > > well, it only took about 2.5 hours to drop the tablespace. Then guess
> > > what, the OS still doesn't see the space. great.
>
> > Also, I forgot to ask. Any idea why I see the contention with inserts
> > while I'm doing a resize or coalesce and I don't see the contention
> > when I'm doing an 'ALTER TABLESPACE ... DROP'?
>
> Looking at wait events will/should tell you what's going on. But what
> could be happening here is that resize /coalesce need to do
> maintenance on fet$/uet$ and that is very expensive if you have lots
> of (small) extents. Drop tablespace will not have the same problem if
> you have previously cleaned up fet$/uet$ by dropping segments and
> resizing down.
When I was seeing other insert statements getting backed up I was
trying to resize. When I did the drop tablespace, the other statements
were not getting backed up. From looking at the top sql during that
period where the tablespace was being dropped there was a delete on fet
$ being ran.
Received on Tue Oct 16 2007 - 08:23:19 CDT
