Rajesh,

Other than the Transaction ID,UBA and Commit SCN (or csc.. Cleanout System
Change)
there is something called 'lock byte' in the ITL entry. THis gives the
details
about the locked (!!) rows. I have dumped few blocks and have an half
written note
with me which explains the lock byte and csc behaviors. I will write to you
offline
after completing the note  once I get some free time.

In a nutshell 'lock byte' and 'csc' is the key here along with the UBA to
get the
read consistent view for the multi update (!) block transactions.

(I have an article in my site which gives an overview about the
transactions, locking
and ITL entry. Have a look there if you have not seen that already!
http://www.geocities.com/kgkrish/transactions.html)

Coming to second  part of your question,

Fixed_tables (views or objects) will not have any information in the
dictionary
and their definitions are coming from the rdbms kernel. So they will not
have a valid
object# or the associating details with them. So that might be the reason
for your
2030 errors (I THink!!)






Best Regards,
K Gopalakrishnan
Bangalore, INDIA



-----Original Message-----
[EMAIL PROTECTED]
Sent: Monday, March 04, 2002 4:53 PM
To: Multiple recipients of list ORACLE-L


Hello Gurus,

As I understand it. In order to provide a read consistent view of the data
while reading a data block, Oracle looks at the SCN in the block header,
and compares it to the snapshot of the SCN taken when the read commenced.
If the Snapshot SCN is less than the SCN in the block header, the query is
directed to read from the rollback segments.

For any transaction that modifies a block, the ITL among other things, also
stores the commit SCN and the address to the transaction table in the
rollback segment. Assume that the block has just one ITL. This ITL can be
reused once the transaction is completed. Assume it is. That is, two
transactions have performed updates on the block since our read commenced.
If so is the case, how does Oracle know which rollback segment to look at?
I am assuming it still looks at the ITL, rolls it back, sees that it needs
to rollback further, looks at the ITL in the rolled back block, and
rollsback further, and so on, until it can reconstruct the data block at an
SCN lower than the snapshot SCN. Is that right?

My second question is, what happens if the data block has two ITL's, both
marked with SCN's greater than when the read commenced. Which ITL does
Oracle look at to get the address of the rollback segment? Is it the one
with the least SCN??

And also, my previous question, remains unanswered so far? Why does Oracle
not allow one to grant select privileges on the fixed tables to any other
user?

Thanks
Raj

--
Please see the official ORACLE-L FAQ: http://www.orafaq.com
--
Author:
  INET: [EMAIL PROTECTED]

Fat City Network Services    -- (858) 538-5051  FAX: (858) 538-5051
San Diego, California        -- Public Internet access / Mailing Lists
--------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).


_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com

-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.com
-- 
Author: K Gopalakrishnan
  INET: [EMAIL PROTECTED]

Fat City Network Services    -- (858) 538-5051  FAX: (858) 538-5051
San Diego, California        -- Public Internet access / Mailing Lists
--------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: [EMAIL PROTECTED] (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).

Reply via email to