Re: Sql*Loader direct slower than nodirect problem...

Giorgio Sorbara <mail_at_fake.org> wrote in message news:<cl0cqm$2rcf$1_at_newsreader1.mclink.it>...

> Performance for 100K records for a direct load is a mess... here
> follows a log:
>
> ----------------------------------------------------------------------
> Record 65504: Rejected - Error on table BIG_TABLE_ENC.
> ORA-00600: internal error code, arguments: [], [], [], [], [], [], [], []

This would be a bug you would have to ask support about...

>
> Record 65505: Rejected - Error on table BIG_TABLE_ENC.
> ORA-00603: ORACLE server session terminated by fatal error
>
> SQL*Loader-926: OCI error while uldlfca:OCIDirPathColArrayLoadStream for
> table BIG_TABLE_ENC

And this might be saying the bug is actually in OCI...

> SQL*Loader-2026: the load was aborted because SQL Loader cannot continue.
> Specify SKIP=65000 when continuing the load.
> SQL*Loader-925: Error while uldlgs: OCIStmtExecute (ptc_hp)
> ORA-03114: not connected to ORACLE
>
> SQL*Loader-925: Error while uldlgs: OCIStmtFetch (ptc_hp)
> ORA-24338: statement handle not executed

Or code that calls OCI. I vaguely recall that old OCI had 64K limits in various places. There are still bugs being introduced into newer releases (like 2763032), maybe you've hit another of those.

jg

--
@home.com is bogus.
http://www.crmbuyer.com/story/37402.html