Re: Oracle 8.16 Deadlock

I'm curious about latch=6. According to v$latchname, it's "longop free list". Why would alter table disable constraint do this? Is there any dependent objects, views or triggers etc, on T49PDGI.PDGI_COUNTRIES that are invalid? I'm not familiar with this latch. But the name tells me there may be something interesting in v$session_longops when the SQL statement is run. Does it run more than 6 seconds?

How about trace the SQL and see what recursive SQLs are being done? What's the session waiting on? And also run alter session set events '4020 trace name errorstack forever' before the alter table?

BTW, Jonathan's DAFT procedue produced ORA-4021, not ORA-4020 for me.

Yong Huang
yong321_at_yahoo.com

"Peter Lasner" <plasner_at_de.imshealth.com> wrote in message news:<3c3ea093_2_at_news2.prserv.net>...
> kkkk-dddd-llll=0000-0005-0005 lock=S pin=S latch=6
Received on Sat Jan 12 2002 - 00:35:09 CST