Re: crash during instance startup with the bad bad ORA-07445 error - suggestions ?
Date: Mon, 31 Dec 2007 08:24:17 +0100
Message-ID: <47789931$0$10626$4fafbaef@reader2.news.tin.it>
"hpuxrac" <johnbhurley_at_sbcglobal.net> ha scritto nel messaggio
news:dd2774fd-d685-4402-b6f9-d66f20df2b58_at_d4g2000prg.googlegroups.com...
On Dec 30, 4:51 am, "KCS" <K..._at_hotmail.it> wrote:
> We have an oracle 8.1.0.5 instance over sun solaris, during a hard user
> deletetion instance crashed, and during startup this is what always
happens
> (down):
>
> From oracle metalink lookup SIGSEGV is a segmentation error.
> What can we do, except calling oracle ?
> The instance version should be desupported.
> Thanks for any suggestion.
>
> Mastino
>
> _=ALERT=_
>
> SMON: enabling cache recovery
> SMON: enabling tx recovery
> Fri Dec 28 11:02:17 2007
> Completed: ALTER DATABASE OPEN
> Fri Dec 28 11:02:17 2007
> Errors in file /oradata1/oracle/admin/****/bdump/****_smon_26257.trc:
> ORA-07445: exception encountered: core dump [kdkkbu()+200] [SIGSEGV]
> [Address no
> t mapped to object] [23808852] [] []
> Fri Dec 28 11:03:14 2007
> PMON: terminating instance due to error 474
> Instance terminated by PMON, pid = 26249
>
> _=TRACE=_
>
> Oracle8i Enterprise Edition Release 8.1.5.1.0 - Production
> With the Partitioning and Java options
> PL/SQL Release 8.1.5.1.0 - Production
> ORACLE_HOME = /oradata1/oracle/product/8.1.5.1
> System name: SunOS
> Node name: ********
> Release: 5.8
> Version: Generic_108528-13
> Machine: sun4u
> Instance name: ********
> Redo thread mounted by this instance: 1
> Oracle process number: 6
> Unix process pid: 26257, image: oracle@********* (SMON)
>
> *** SESSION ID:(5.1) 2007.12.28.11.02.17.000
> Exception signal: 11 (SIGSEGV), code: 1 (Address not mapped to object),
> addr: 0x
> 16b4b54, PC: kdkkbu()+200
> *** 2007.12.28.11.02.17.000
> ksedmp: internal or fatal error
> ORA-07445: exception encountered: core dump [kdkkbu()+200] [SIGSEGV]
> [Address not mapped to object] [23808852] [] []
> ----- Call Stack Trace -----
> calling call entry argument values in hex
> location type point (? means dubious value)
> -------------------- -------- -------------------- -----------------------
--Received on Mon Dec 31 2007 - 01:24:17 CST
> ---
> ksedmp()+160 CALL ksedst()+0 A8 ? 1 ? FFBEB544 ?
> FFBEAFE8 ? FFBEAFCC ? 0 ?
> ssexhd()+368 CALL ksedmp()+0 3 ? 0 ? 1 ? FFBEBA5C ? 1
> ?
> 7 ?
> sigacthandler()+40 PTR_CALL 00000000 B ? 1634400 ? FFBEBDF0 ?
> 1634400 ? 0 ? 0 ?
> ----- Argument/Register Address Dump -----
> Argument/Register addr=ffbeb544. Dump of memory from 0xFFBEB504 to
> 0xFFBEB644
> FFBEB500 726E616C 206F7220 66617461 6C206572 726F720A 00BEB980
> 00111170
> FFBEB520 1200000A 0008E4AA 0162D860 00000000 12000000 0008E4B3 00414284
> 0000007
Do you have metalink access? This can be researched inside oracle support if you do. Is there left over shared memory still allocated to the crashed instance in the os? You can use ipcs to check this out and ditch the left over memory segments if applicable. __________ We have already done it. Thanx.