Re: semophore errors in Oracle 6.

From: Herbert van den Bergh <hbergh_at_nlsu10.nl.oracle.com>
Date: Fri, 27 May 1994 10:12:15 GMT
Message-ID: <HBERGH.94May27121215_at_nlsu10.nl.oracle.com>


In article <dmausner.3.000AA0C9_at_brauntech.com> dmausner_at_brauntech.com (Dave Mausner) writes:

   In article <CqA9Cu.8H6_at_ssigv.UUCP> thongj_at_tus.ssi1.com (James Hong) writes:
>From: thongj_at_tus.ssi1.com (James Hong)
>Subject: semophore errors in Oracle 6.
>Date: Tue, 24 May 1994 01:26:54 GMT
 

>Dear Netter
 

>Frequently, I am getting following error message in alert log file.
 

>Fri May 20 13:48:02 1994
>Error signalled in file: /usr/bin/oracle/rdbms/log/13_27234.trc
>ORA-07265: sppst: semop error, unable to increment semaphore.

   I have seen this, too, under SCO. The answer was to reduce the potential    number of users in the INIT.ORA script. The maximum number of PROCESSES    defines the limit on the semaphore count, or something like that. We knocked    the maximum user/process count down from 60 to 25 and sneaked under the limit.

   otherwise, we would have required a re-link of the unix kernel: yeccch.    hope the above helps out.

   Dave Mausner, Sr Consultant, Braun Technology Group, Chicago.

I believe the cause of the messages the original poster finds in his alert file is that the Oracle instance was killed by a shutdown abort, while users were still connected to the database. The oracle shadow processes write the message to the alert file when they try to operate on a semaphore which is gone after the shutdown abort. Changing init.ora parameters will not affect this.

The error message that is the result of starting a database with too few semaphores configured in the system is written to the terminal, and is probably an ORA-07252 error, not ORA-07265.

Herbert. Received on Fri May 27 1994 - 12:12:15 CEST

Original text of this message