Re: A new form of ORA-1555

2003-06-07 Thread Mladen Gogala
I'm not using AUM on my production databases because they're still 8.1.7. To tell the truth, I haven't seen any major patches for AUM in the 9iR2. Well, when Oxford goes to 9.2, I guess AUM is an option. I'm still paranoid about the new features. I believe that progress does bad things to the

Re: A new form of ORA-1555

2003-06-06 Thread Kirtikumar Deshpande
A classic case when AUM does not help prevent 1555 errors ! Query duration is MAXQUERYLENreported inv$undostat view. SCN could be the 'as of SCN' when the query started (not sure, as I could never get my small tests to failwith 1555 when using AUM). What is also interesting is the SMON

Re: A new form of ORA-1555

2003-06-06 Thread Rachel Carmichael
we ended up with an ORA-600 after lots of 'SMON offlining undo segment messages in the alert log. Happened only once, Support asked us to set an event and trap it when/if it happens again I'm beginning to think maybe I should have stuck with rollback segments --- Kirtikumar Deshpande [EMAIL

Re: A new form of ORA-1555

2003-06-06 Thread Ruth Gramolini
Title: A new form of ORA-1555 It might have something to do with setting the table to do a Flashback query. Maybe Dan will know. Ruth - Original Message - From: Jamadagni, Rajendra To: Multiple recipients of list ORACLE-L Sent: Friday, June 06, 2003 10:00 AM

Re: A new form of ORA-1555

2003-06-06 Thread Daniel W. Fink
You do not have to specifically enable a table for flashback query. In 9.2 (Kirti, please correct me if I am wrong), it is automatically enabled, even if you are not using AUM. I believe that this is a new error message format for 9. The solution that is commonly quoted is to increase your

RE: A new form of ORA-1555

2003-06-06 Thread Gogala, Mladen
Flashback query is enabled by setting the UNDO_RETENTION parameter to the desired number of seconds. Table V$UNDOSTAT contains the column MAXQUERYLEN which gives the max time that you can use to dynamically set UNDO_RETENTION. You also have to enable flashback for session using

Re: A new form of ORA-1555

2003-06-06 Thread Rachel Carmichael
well it's been a few weeks and ours are still there, offline so I guess they don't get dropped. --- Daniel W. Fink [EMAIL PROTECTED] wrote: You do not have to specifically enable a table for flashback query. In 9.2 (Kirti, please correct me if I am wrong), it is automatically enabled,

Re: A new form of ORA-1555

2003-06-06 Thread Kirtikumar Deshpande
Acutally, there nothing at the table level to 'enable' it for the FBQ (in 9i R1 and R2). In 9i R2, they introduced a 'flashback' privilege (object and System level). In 9i R1 one needed exec privilege on a package (dbms_flashback) to use FBQ. SMON offlined Undo Segs because those had been

Re: A new form of ORA-1555

2003-06-06 Thread Daniel W. Fink
Mladen, That is a common misconception, one that Oracle wholeheartedly supports. Flashback Query (FBQ) depends upon the smon_scn_time table, which is popluated in 9i, regardless of the undo_management setting. This table is populated regardless of the UNDO_RETENTION parameter. This parameter

RE: A new form of ORA-1555

2003-06-06 Thread Gogala, Mladen
Well, I've read a lot of that on this list (this is not the first time FBQ is being discussed) but as a conservative and somewhat paranoid DBA, I don't want to try anything that isn't supported with a very new feature like FBQ. The experience taught me a lesson about ora-600 and alike. I

Re: A new form of ORA-1555

2003-06-06 Thread Jared . Still
] Sent by: [EMAIL PROTECTED] 06/06/2003 03:20 PM Please respond to ORACLE-L To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED] cc: Subject:Re: A new form of ORA-1555 Mladen, That is a common misconception, one that Oracle wholeheartedly

Re: A new form of ORA-1555

2003-06-06 Thread Daniel W. Fink
Mladen, I could not agree more! I seriously pondered not posting this information at all. FBQ is a nice feature, but I would not depend upon it. I'm a conservative and somewhat paranoid DBA and I would not recommend AUM for production systems, though certain very knowledgable and respected

Re: A new form of ORA-1555

2003-06-06 Thread Rachel Carmichael
we are using AUM pretty successfully for our production systems. Interestingly enough the one problem we did have was on a test box, and we couldn't repeat it --- Daniel W. Fink [EMAIL PROTECTED] wrote: Mladen, I could not agree more! I seriously pondered not posting this information at

Re: A new form of ORA-1555

2003-06-06 Thread Mladen Gogala
Oh no, we agree. I wouldn't do automatic undo management, either. V$ROLLSTAT is perfectly good for me. I'll wait for the version 10 to intorduce AUM to my databases. On 2003.06.06 21:54 Daniel W. Fink wrote: Mladen, I could not agree more! I seriously pondered not posting this

Re: A new form of ORA-1555

2003-06-06 Thread Kirtikumar Deshpande
I am using AUM in our Test/Acceptance databases with no problems at all. This month it will be rolled out to a couple of production databases. FBQ is another matter altogether :) MUM (manual undo mgmt) was a depreacted option when 9i R1 came out. I won't be surprised if only AUM would be