RE: Undo Usage and Read consistency - ORA-1555

From: Ric Van Dyke <ric.van.dyke_at_hotsos.com>
Date: Tue, 14 Jul 2009 14:16:28 -0500
Message-ID: <C970F08BBE1E164AA8063E01502A71CF014E0E76_at_WIN02.hotsos.com>



That would only if you can do the COMMIT in the process that is holding the REDO.  

Doing a COMMIT in the session that is doing the SELECT will have no impact on 1555's. To solve this is an application design or usage issue. It's very likely that a user terminating in another session could be the cause.  


Ric Van Dyke

Hotsos Enterprises


 

Hotsos Symposium

March 7 - 11, 2010

Be there.    


From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Andrew Kerber Sent: Tuesday, July 14, 2009 11:53 AM
To: nancy_iles_at_hotmail.com
Cc: cicciuxdba_at_gmail.com; oracle-l_at_freelists.org Subject: Re: Undo Usage and Read consistency - ORA-1555  

When a short select causes a ora-1555, it usually means that that select is part of a larger transaction which began earlier than data is available in the undo. One simple method would be to issue a commit prior to doing the select, if that is possible. If it is not possible, you may need to increase the undo retention.

On Tue, Jul 14, 2009 at 10:35 AM, Nancy Iles <nancy_iles_at_hotmail.com> wrote:

We have sporadic ora-1555 on an exceptionally simple statement that occurs frequently in the application. How can you analyze why this tiny, simple statement is causing an ORA-1555? The statement is:

SELECT RESV_NAME_ID , RESORT FROM RESERVATION_NAME WHERE CONFIRMATION_NO = :1 I believe that this causes a user session to hang. Our third party vendor says that it is because a user terminates their session improperly and that it is not an issue.

Any suggestions on how to analyze the cause and the system impact?

TIA, Nancy Iles
Omni Hotels



> Date: Tue, 14 Jul 2009 10:24:00 -0300
> Subject: Undo Usage and Read consistency - ORA-1555
> From: cicciuxdba_at_gmail.com
> To: oracle-l_at_freelists.org
>
> Dear DBAs,
>
> We got into a discussion about how read consistency is implemented in
Oracle and was wondering what you know of this.
>
> The two sides are the following:
> 1) Undo is used for any and all selects, so if you do a full scan on a
large table it is bound to give ORA-1555 even if there is no transaction modifying the table
>
> 2) Undo is only used when a transaction modifies data (DML) and ONLY
then ORA-1555 is possible, since it happens when the consistent version of the block stored in the UNDO by the transaction ages out.

>
> Documentation is unclear to this respect:
>
> From the concepts guide:
>

> To manage the multiversion consistency model, Oracle must create a
> read-consistent set of data when a table is queried (read) and
> simultaneously updated (written). When an update occurs, the original
> data values changed by the update are recorded in the database undo
> records. As long as this update remains part of an uncommitted
> transaction, any user that later queries the modified data views the
> original data values. Oracle uses current information in the system
> global area and information in the undo records to construct a
read-consistent view of a table's data for a query.
>
> Does this mean that every time I perform a select I get a copy of the
data into de undo?
>
>
> Alan Bort
> Oracle Certified Professional



Hotmail(r) has ever-growing storage! Don't worry about storage limits. http://windowslive.com/Tutorial/Hotmail/Storage?ocid=TXT_TAGLM_WL_HM_Tut orial_Storage_062009--
http://www.freelists.org/webpage/oracle-l
-- 
Andrew W. Kerber

'If at first you dont succeed, dont take up skydiving.'


--
http://www.freelists.org/webpage/oracle-l
Received on Tue Jul 14 2009 - 14:16:28 CDT

Original text of this message