Re: Deadlock inserting into same rowid (different block)

  • From: "Jonathan Lewis" <jonathan@xxxxxxxxxxxxxxxxxx>
  • To: <oracle-l@xxxxxxxxxxxxx>
  • Date: Wed, 20 Mar 2013 18:44:45 -0000

The "graph" isn't pointing to any object.

The rowids waited on reference an object - which I think you're saying is 
the index supporting the foreign key - but I've said the unlikely 
occurrence of the zeros as both slots numbers is, by itself, sufficient to 
make you suspicious that the reporting is false. (The fact that the object 
is an index, even more so).

I'd hazard a guess that you'll find that that index is a popular candidate 
for "buffer busy wait" waits (check v$segstat) and that the inserts are 
(roughly) sequential on the parent id.


Regards

Jonathan Lewis
http://jonathanlewis.wordpress.com/all-postings

Author: Oracle Core (Apress 2011)
http://www.apress.com/9781430239543

----- Original Message ----- 
From: "Patterson, Joel" <jpatterson@xxxxxxxxxx>
To: "Bobak, Mark" <Mark.Bobak@xxxxxxxxxxxx>; "mohamed houri" 
<mohamed.houri@xxxxxxxxx>
Cc: "Jonathan Lewis" <jonathan@xxxxxxxxxxxxxxxxxx>; 
<oracle-l@xxxxxxxxxxxxx>
Sent: Wednesday, March 20, 2013 6:05 PM
Subject: RE: Deadlock inserting into same rowid (different block)


Ok, I get the deadlock scenario.   What has thrown me off is that the graph 
is pointing to the object FK Index...  as opposed to the PK index or parent 
row.   Since there can be several child entries for the same parent.

But what to say now as a suggestion.  Therefore just say to stand by and 
see if it becomes a recurring theme?

Can it be an anomaly seeing how there was a suggestion "... Given that the 
slot numbers are both zero, it's likely that they're being reported 
incorrectly... "

It is the same app for both sessions so it is inserting the tables in the 
same order...

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

Other related posts:

  1. Home
  2. oracle-l
  3. Archive
  4. 03-2013