Re: Anyone dealt with address not mapped to object errors?

Are you using any of the following in the problem SQL:

    parallel query
    partitioned tables
    partition views
    global temporary tables
    partitioned index organized tables
    index fast full scan optimization paths

--

Jonathan Lewis
Yet another Oracle-related web site:  http://www.jlcomp.demon.co.uk

Dave A wrote in message ...


>Hi all,

>

>Of the many databases I support under Oracle 7.2.3-8.1.6.2 only one

 database


>is currently giving me grief that I can't get a handle on.

>I should mention that I can reproduce the problem at will on our production

>server and not anywhere else under any Oracle version.

>

>I have Oracle support working on the issue, but it's been a week and they

>have nothing to tell me yet.

>

>The database version is 8.1.6.2 recently upgraded from 7.3.4.3 with the

>migration utility.  It runs on a sun 6500, 10 processors (336mhz, 10 gb ram

>solaris 2.6 with all required patches)

>

>The problem is that the database is core dumping about 20-30 times per day

>(doesn't crash).  The trace files reveal 7445 errors (address not mapped to

>object) followed by a 600 error.  There seem to be about 6 or so distinct

>sql statements in memory during the error(one statement per trace file/core

>dump).  Any one of these statements can cause the error.

>

>I have run these statements from sqlplus and cannot reproduce the problem.

>Today, however, one of my own statements blew up with a 3113 error(end of

>file on communication channel).  Sure enough, there was a core dump and

>trace file with the same errors as all the others.  The only peculiar thing

>about the statement I was running is that it was a piece of dynamic sql

 that


>had a syntax error in it.  Under any other DB the error is "from clause not

>found where expected"(or something similar).  On this one DB though and end

>of file on communication channel error occurs.  I can reproduce this error

>at will by running this statement.

>

>I set up a test database where I migrated from 7.3.4 to 8.1.6.2 using the

>same data our production system had.  The init.ora parameters are identical

>except lesser block buffers and one less rollback segment.  I cannot

>reproduce the error there.

>

>Other tidbits are that the application causing these errors connects to the

>DB via SQLNet and my SQLPlus session was a direct connect so SQLNet/network

>issues don't appear to be an issue.

>

>I suspect the 3113 error is being caused by the 7445 and 600 errors

>disconnecting the process from the database.

>

>All the sql encountering this problem are select statements, no dml.(so

 far)


>

>So, why am I telling you all this?  I am hoping that per chance one of you

>have encountered something similar that might help me get a handle on the

>situation.

>

>TIA,

>--

>Dave A

>

>

>

>