HPE Community read-only access December 15, 2018
This is a maintenance upgrade. You will be able to read articles and posts, but not post or reply.
Hours:
Dec 15, 4:00 am to 10:00 am UTC
Dec 14, 10:00 pm CST to Dec 15, 4:00 am CST
Dec 14, 8:00 pm PST to Dec 15, 2:00 am PST
Operating System - OpenVMS
cancel
Showing results for 
Search instead for 
Did you mean: 

Oracle Export Crash

 
Niall Godwin
Advisor

Oracle Export Crash

We run Oracle V7.3 on an OpenVMS AXP V7.1-2 server.
An Oracle (Complete) Export crashed on two occasions for the following reasons:

Crash #1:
%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=00000000010A7FC0, PC=00000000002115C4, PS=0000001B

Crash #2:
%SYSTEM-F-ASTFLT, AST fault, SP=010A7FC0, param=00000BAD, PC=0000000000211688, PS=0000001B, target PC=0000000000000BAD, PS=0000001B

Exports on other occasions executed without error.

Any help would be appreciated.

Thank You
Niall
4 REPLIES
Marc Van den Broeck
Trusted Contributor

Re: Oracle Export Crash

Hi,

does a previous version of the filename you use in the export exists?
If so, the newly created export file takes some attributes of the 'old' file and the export may fail.
This happened to me some time ago and it took me some time to find out why.

Rgds
Marc
Volker Halle
Honored Contributor

Re: Oracle Export Crash

Niall,

these are software programming errors, most likley in the Oracle code. The program counter in both cases is within P0 space, the user program space. If you have a process running the same image, you could use SDA> SHOW PROC/IMA to find out, which code is loaded at that address range.

To collect more information, in case this happens again, you could activate process dumps before running these images:

$ SET PROC/DUMP

then run the image. Or - if it's a detached process and you can influence the RUN command, add a /DUMP.

In case of an improperly handled condition, a process dump would be generated in the default directory and called imagename.DMP. You can analyze such a dump with ANAL/PROCESS.

If you've collected a process dump and have a service contract with Oracle (assuming that V7.3 is still a supported version), you could log a call. Unfortunately, process dump analysis on OpenVMS V7.1-2 process dumps is - in most cases - bound to the system, where the process dump has happenend.

Consider to obtain and install the most recent available patches for Oracle.

Volker.
Hein van den Heuvel
Honored Contributor

Re: Oracle Export Crash

I like Martin's explanation. This is a relatively recent change to the C-RTL (for a ntopic talking ancient stuff like Oracle V7.3 and VMS 7.1). To inherit certian file attribute which then later may bite you, and may bite an existing program like exp.

That said, I also agree with Volker (That's a no-brainer.. he is always right about VMS thingies :-):
"these are software programming errors, most likley in the Oracle code."

Specially crash #2. Check out that 'target PC' and 'param'... The HEX address actually reads "BAD". So some code (EXP? VMS?anticipated it could get in trouble there!



May we assume that the basic setup has been running succesfully for years and years, exporting (as backup?) frequently? What changed?
- DB size?
- Disk space?
- OS upgrade/patch?
- Hardware (cpus, mem, io)
- Was there an ORA-006xx error reported?
- Was there a .TRC file generated
- Is it related to a specific TABLE/USER?


Even if nothing changed you'll need to capture a bunch of this before raising a call to Oracle.
Of course they might just laugh when you mention 7.3.. and VMS :-(.

Hein.
Niall Godwin
Advisor

Re: Oracle Export Crash

Marc,
The previous (export) file is deleted before beginning the export.

Volker,
Yes, worth trying.

Hein,
No messages in the ALERT log.
The crashes occurred when exporting different tables.
No changes to software or hardware.

Thanks to all for your suggestions.
I will put the question to some of the specialist Oracle fora.

Niall.