Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!sun-barr!cs.utexas.edu!know!zaphod.mps.ohio-state.edu!samsung!crackers!cpoint!frog!barr
From: barr@frog.UUCP (Chris Barr)
Newsgroups: comp.databases
Subject: Re: Oracle crashes
Summary: For unix: delete sgadef{SID}.dbf before restart; use 2 controlfiles.
Keywords: help oracle startup shutdown
Message-ID: <19503@frog.UUCP>
Date: 4 Oct 90 00:58:00 GMT
References: <25244@sequoia.execu.com>
Organization: Charles River Data Systems
Lines: 25

In article <25244@sequoia.execu.com>, bob@sequoia.execu.com (Bob Kieras) writes:
> 
>     We are running Ultrix-32 V3.1 (Rev. 9) on a Microvax 3600 and
>     ORACLE RDBMS V6.0.27.9.1 and we have encountered a problem.
>     If the machine is shutdouwn using the shutdown command without
>     shutting oracle down first, the database cannot be restarted. 
> 	 Unexpected oracle server termination by signal 6
>     ( Core is dumped and oracle crashed.)

Under Unix, a different message appears at restart, but the solution is
to first delete the ORACLE_HOME/dbs/sgadef{SID}.dbf file, then startup
will recover.  This might work for VMS - it works under Unix even when
the plug was pulled.

>     I have traced the problem to the database control file, 
>     /usr/oracle/dbs/cntrlora6.dbf where "ora6" is the instance name.
>     This file is either broken or left in a state incompatible with 
>     restarting the database. 

Use at least 2 controlfiles, specified at create time and in init.ora.
They are kept identical - except in case of disk crash or, perhaps, your
poweroff, in which case you'll have two to explore ...

Chris Barr
!frog!barr