Re: Help with ORA-08176 please

Wait a minute
I`m not fully awake and did not read you well

> > Redo is not used in the creation of consistent read blocks.
> > Undo is used in the creation of consistent read blocks.

You are right.
Now I understand and all is clear but:

Is that to say that ORA-8176 has the same cause as ORA-1555 ? Why 2 messages then?
Is there something fundametally different in the 2 things?

Thanks.

-- 
Syltrem
OpenVMS 7.3-1 - Oracle 8.1.6.0

http://pages.infinit.net/syltrem (OpenVMS site in french language)
"Syltrem" <syltremzulu_at_videotron.ca> a écrit dans le message de
news:2tcfb.39077$1M6.494192_at_wagner.videotron.net...


> > <no Oracle Server version info posted>

> Look at the end - it is: OpenVMS 7.3-1 - Oracle 8.1.6.0

>

> > NOLOGGING does not affect the creation of undo (rollback).

> agreed

>

> > NOLOGGING affects the generation of redo.

> agreed

>

> > Redo is not used in the creation of consistent read blocks.

> > Undo is used in the creation of consistent read blocks.

> don`t agree. *Rollback segments* are used for consistent view.

>

> Pls correct me if I`m wrong

> I still don't understand what caused ORA-8176

>

> Thanks

>

> -- 

> Syltrem

> OpenVMS 7.3-1 - Oracle 8.1.7.4

>

> http://pages.infinit.net/syltrem (OpenVMS site in french language)

> "Paul Drake" <drak0nian_at_yahoo.com> a écrit dans le message de

> news:1ac7c7b3.0310022215.a580aa6_at_posting.google.com...

> > "Syltrem" <syltremzulu_at_videotron.ca> wrote in message

> news:<s0_eb.2547$G1.12138_at_tor-nn1.netcom.ca>...

> > > Hi

> > >

> > > Yesterday a report was running for an hour or so, when the user called

> and

> > > asked why it was taking so long.

> > > I found I needed to create an index so I did.

> > >

> > > CREATE INDEX .... NOLOGGING;

> > > ALTER INDEX ... LOGGING;

> > >

> > > I thought creating the index with NOLOGGING would be a good thing

since


> this

> > > was a pretty large table, and the index creation would run faster.

> > >

> > > But the report program stopped abruptly at some point with the message

> > >

> > > 08176, 00000, consistent read failure; rollback data not available

> > >

> > > // *Cause:   Encountered data changed by an operation that does not

> generate

> > > //           rollback data : create index, direct load or discrete

> > > transaction.

> > > // *Action:  In read/write transactions, retry the intended operation.

> Read

> > > //           only transactions must be restarted.

> > >

> > >

> > > Can you explain why a READ transaction would suffer from this (CREATE

> > > INDEX... NOLOGGING) ?

> > > To me, Oracle should just start using the index next time it executes

> the

> > > problem request again, following index creation. The request that

> benefitted

> > > from the new index is executed hundreds of time in that report.

> > >

> > > After restarting the report program it took only 2 minutes to run.

> > >

> > > Thanks

> >

> > Hi.

> >

> > <no Oracle Server version info posted>

> > NOLOGGING does not affect the creation of undo (rollback).

> > NOLOGGING affects the generation of redo.

> > Redo is not used in the creation of consistent read blocks.

> > Undo is used in the creation of consistent read blocks.

> > More than likely, the rate of undo generation caused blocks of undo

> > that were required to generate a consistend read (at that point) to be

> > over-written (previously).

> >

> > might I suggest having a look at some papers on this topic, here's a

> > good start:

> >

> > http://www.nyoug.org/techjournal.htm

> >

> > click on Tim Gorman's presentation.

> >

> > hth.

> >

> > Paul

>

>