Re: deadlock detection/resolving

It did resolve it automatically. It found the deadlock within three seconds, rolled back one of the transaction AT THE STATEMENT LEVEL (which is the only safe rollback it could do) and sent the transaction an error message that # it had caused a deadlock so that the calling program would know it had to take appropriate action.

Since the appropriate action is ENTIRELY dependent on the prior activity of the calling program, it would be dangerous for any further action on the part of the database.

Put the question another way - what more would you want it to do ? (And would this make sense in the general case, rather than your particular situation ?)

--
Jonathan Lewis
http://www.jlcomp.demon.co.uk

Author of:
Practical Oracle 8i: Building Efficient Databases

Next Seminar - Australia - July/August
http://www.jlcomp.demon.co.uk/seminar.html

Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html



Rüdiger J. Schulz wrote in message <3cc02889_at_netnews.web.de>...


>hi all,

>

>

>my oracle-server (8.1.7 on openunix 8/lkp) detects a dealock

>(ora-000060 in the alert.log / and the trace-file).

>the situation: two statements are trying an UPDATE on two rows.

>

>

>but the oracle-server don't resolve it automatic!

>

>

>why?

>