Re: ORA-08177 at insert?
- From: Martin Berger <martin.a.berger@xxxxxxxxx>
- To: Toon Koppelaars <toon.koppelaars@xxxxxxxxxxx>
- Date: Fri, 21 Jun 2013 13:24:12 +0200
Toon,
thank you for your comment.
there IS an index on the table (and without an index - the problem does not
reproduce at all).
It somehow makes sense as there is an 'update' in the index structure.
Now you directed my curiosity away from blocks to index splits - and how
uncommitted entries are handled exactly (and why they harm serializable or
are secured by rowdependencies).
Martin
On Fri, Jun 21, 2013 at 1:09 PM, Toon Koppelaars <
toon.koppelaars@xxxxxxxxxxx> wrote:
> Is there an index on that table?
>
> Inserts cause index "updates". Many years ago I investigated what might
> have been the exact same issue you describe here, and I recall that as soon
> as index maintenance decides that a block needed to be split, you could run
> into the 'cannot serialize' error.
>
>
> On Fri, Jun 21, 2013 at 12:21 PM, Martin Berger <martin.a.berger@xxxxxxxxx
> > wrote:
>
>> Hi List,
>> I found a very interesting observation I can not explain (based on this
>> page:
>> http://www.devx.com/dbzone/Article/41591/0/page/2 )
>>
>> given a table with an index (not even unique)
>>
>> with 2 sessions set isolation_level=serializable and both do a simple
>> insert often enough, an
>> ORA-08177: can't serialize access for this transaction
>> is raised.
>> The solution of the page mentioned above is to set the table to
>> rowdependencies
>> .
>>
>> But I would like to understand why the ORA-08177 is raised at all for pure
>> inserts? They should be totally independent?
>>
>>
>
>
> --
> Toon Koppelaars
> RuleGen BV
> Toon.Koppelaars@xxxxxxxxxxx
> www.RuleGen.com
> TheHelsinkiDeclaration.blogspot.com
>
> (co)Author: "Applied Mathematics for Database Professionals"
> www.rulegen.com/am4dp-backcover-text
>
--
//www.freelists.org/webpage/oracle-l
Other related posts:
