Re: Import ora-1658 initial extent size

You have probably done the export with compress=y (which is the default). This will create one large initial extent for each segment (ie. all extents are compressed into one large initial extent). The reason you see the error is that one extent cannot be divided across multiple datafiles.

There are three solutions (that I can think of right now):

1. create larger datafiles (large enough to hold the initial extents) - 4Gb is the max on NT as far as I remember.
2. run the export again, this time explicitely specifying "compress=n" on the exp80 command line
3. create all tables and indexes with the preferred storage clauses (use indexfile option on import to get DDL's for all tables and indexes) and then import with "ignore=y".

Obviously (1) is the easiest way, (2) is second-best and should be used if one of the initial extents are larger than 4Gb, but if you do not have access to the database that were exported then (3) might be the only way. A combination of (1) and (3) could also be used if you only have a few segments with very large initial extents.

HTH,
Gollum

"Peter Laursen" <ptl_at_edbgruppen.dk> wrote in message news:97l2vp$nnu$1_at_news.inet.tele.dk...
> Hi,
>
> System: Oracle 8.0.5.2.1/ NT 4.0
> I have an export file done as user sys with full=y and all other
 parameters
> default, ie compress=y conventional path.
> I need to import this into to an empty database with precreated
 tablespaces.
> The import must create the users and system data like roles and synonyms.
> The easiest way was to do imp as system with full=y, but I run into
> ora-1658 cannot create initial extent for ... the initial extents given in
> the imp file are to big to fit in the 1G datafiles.
>
> What is the best and most efficient way to import from the given
 exportfile?
>
> TIA
> Peter Laursen
>
>
>
>
Received on Thu Mar 01 2001 - 04:12:18 CST