Re: ORA-01115 IO errors

From: Gerald Sinkiewicz <sinkiege_at_snet.net>
Date: Sun, 1 Jul 2007 20:17:52 -0400
Message-ID: <QuXhi.8538$bP5.7165@newssvr19.news.prodigy.net>

"DA Morgan" <damorgan_at_psoug.org> wrote in message news:1183301755.616247_at_bubbleator.drizzle.com...

> Gerald Sinkiewicz wrote:
>> "DA Morgan" <damorgan_at_psoug.org> wrote in message 
>> news:1183262863.121832_at_bubbleator.drizzle.com...

>>> Gerald Sinkiewicz wrote:
>>>
>>>> The ORA-01115 references file 10. I cannot post file names.
>>>> Ah, I did misspell "owner" and there are other messages related to the
>>>> ORA-01115 that really just say Oracle could not read a certain block in
>>>> file 10....
>>> That is an entirely different matter that has nothing to do with what
>>> you sent the storage admins chasing ... stop them.
>>>
>>> Oracle error messages are intended to be handled in a specific manner as
>>> befits an enterprise database. First you go to
>>> http://metalink.oracle.com and determine an appropriate actions plan:
>>> Then you act precisely as instructed. What you appear to have done is
>>> a guaranteed recipe for disaster ... not understand the message, not
>>> research the message ... and start doing stuff hoping something will
>>> fix it.
>>>
>>> So lets get back to basics ...
>>>
>>> Oracle can not read one or more specific blocks of a specific data file.
>>> This is an indication of a specific problem with a specific solution.
>>> Which blocks? In which tablespace? Being used by which objects? What
>>> version of Oracle do you have and have you, as you should have, taken
>>> backups with RMAN? If so what backups do you have, how were they taken,
>>> and specifically what actions, if any, have been taken since the problem
>>> was first discovered.
>>>
>>> While you are contemplating the above run an RDA and open an SR.
>>> --
>>> Daniel A. Morgan
>>> University of Washington
>>> damorgan_at_x.washington.edu (replace x with u to respond)
>>> Puget Sound Oracle Users Group
>>> www.psoug.org

>>
>> Of course, I did go to metalink, which directs me to see the storage 
>> guys.
>> Since the errors of 6/27 the problem has gone away, that is the 
>> application is no longer getting the error.
>> I have dbv'd the files in question and my backups are also verified and 
>> do not point to a problem.
>> The problem is intermittent, which is one reason metalink sends me to the 
>> guys with the McData's.
>>
>> If the problem had shown up in my alert log I would have been able to 
>> submit an SR to oracle, but since it is
>> in application logs that I cannot see, there is not much I can send in 
>> with the SR for oracle to look at other than
>> an RDA. I really think the SAN guys need to chase something here because 
>> there is a history in errpt of SAN
>> disk errors. It is possible the McData switch has intermittent problems, 
>> that could cause some strange errors, like
>> this one. It is also possible there is a server problem I suppose, like 
>> memory, that could mess up a bit and cause
>> a strange error. The only thing I am sure of is file 10 is not corrupt 
>> :-).
>>
>> But are there any references for "RISC System/6000 Error:1: Not owner"? I 
>> did request the system admins to
>> contact IBM about the meaning, but have not heard a word, yet.
>
> Given the clarification ... I'd be looking at hard disk controller cards
> on the server and related hardware infrastructure on the SAN. Something
> may be getting prepared to die.
> -- 
> Daniel A. Morgan
> University of Washington
> damorgan_at_x.washington.edu (replace x with u to respond)
> Puget Sound Oracle Users Group
> www.psoug.org

I will agree with that, this box has not been well. In the recent past this box has died at least once (bad power supply is what IBM said)
and hung a couple of times (once running out of pagefile space). Received on Sun Jul 01 2007 - 19:17:52 CDT