Note in-line
--
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/seminar.html
Optimising Oracle Seminar - schedule updated July 27th
"Ana C. Dent" <anacedent_at_hotmail.com> wrote in message
news:Xns9556BE75F5D1DSunnySD_at_68.6.19.6...
> netcomradeNSPAM_at_bookexchange.net (NetComrade) wrote in
> news:4134e977.528116261_at_localhost:
>
>
> Keep in mind that a LOCK is not the same as an ENQUEUE (wait);
> otherwise they'd have the same name & not be named differently.
>
The sentence sounds logical - however it is not true.
The types of things you see in v$lock are exactly the types of
things you see in v$enqueue_stat (v9), the things reported under
"enqueue" statistics in v$sesstat and v$sesstat and under the
"enqueue" waits in v$session_wait et. al.
A lock is an enqueue is a thing hanging off one of the
queues on an enqueue-resource
> ENQUE wait are typically caused by a combination of full data blocks
> and INITRANS being "too small". When a block is "too full" the
> Interested Transaction List (ITL) is precluded from growing to
> hit MAXTRANS. So any transaction which desires to modify a block,
> such as an Index Header block, must wait for an available slot
> in the ITL.
The example you describe happens to be one reason for an
enqueue/lock going onto the waiters' queue on an enqueue_resource
of type TX in mode 4 - typically whilst another process has an
enqueue/lock on the holders' queue on the same enqueue_resource.
Received on Wed Sep 01 2004 - 14:42:40 CDT
