Re: sqllder hangs at 99-100% CPU on select of fet$

Hi!

Last time I saw a similar problem, the client who was modifying data was stuck behind locks on fet$ & uet$ which were stuck behing SMON which was stuck behind parallel slaves trying to do parallel recovery to a failed long transaction. Unfortunately I don't think (or know) that there are parallel recovery capabilities in Oracle 7, but you might want to check what SMON is doing there. Is it recovering something or pehaps coalescing space or getting rid of temp segments in temporary (or some other) tablespace.

2 cents,
Tanel.

"Don Seiler" <don_at_seiler.us> wrote in message news:f6r7b.487$ev2.331412_at_newssrv26.news.prodigy.com...
> I made an earlier post but the info in that is incorrect and I can't
> wait for it to show up on usenet.
>
> Using Oracle 7.3.4.5 on HP-UX. I have an application that uses sqlldr
> to bulk load a table. The table usually has around 65 million rows.
> This application has worked fine in the past, but for some reason this
> week it isn't working. When the application is started, it might load
> some (once it got up to 90,000 records, now it does 0), but eventually
> it will hang, the oraclesid process for it will go to 99-100% CPU usage,
> and nothing will happen.
>
> I load up TOra to see what it is doing and this is the current SQL it
> shows me for that process:
>
> SELECT length
> FROM fet$
> WHERE file# = :1
> AND block# = :2
> AND ts# = :3
>
> I'm assuming that sqlldr does that on it's own because my developers
> have no idea what fet$ is. This process locks fet$ and eventually will
> block SMON from accessing the table. This whole mess makes the database
> unusable until I kill the sqlldr's oraclesid process.
>
> Any remedies would be welcomed with open arms.
>
Received on Tue Sep 09 2003 - 17:58:03 CDT