Re: RMAN not retrying on media manager errors
From: Robert Klemme <shortcutter_at_googlemail.com>
Date: Wed, 07 Apr 2010 20:41:39 +0200
Message-ID: <8241v3FuddU1_at_mid.individual.net>
On 06.04.2010 16:00, Ian Chard wrote:
> I'm using Oracle 10g, RMAN and TDPO (the TSM client for Oracle). I'm
> trying to do a PITR to a non-production machine, but RMAN is intolerant
> of any media manager problems that crop up, so things like this
>
> ORA-19511: Error received from media manager layer, error text:
> ANS1017E (RC-50) Session rejected: TCP/IP connection failure
>
> and this
>
> ORA-19511: Error received from media manager layer, error text:
> ANS1314E (RC14) File data currently unavailable on server
>
> result in an immediate 'failover to previous backup'. The first error
> was caused by a transient network problem; I suspect the second was just
> bad luck as the file was being reclaimed by the TSM server when TDPO
> asked for it.
>
> Both these errors would have gone away if RMAN had tried again, so is
> there any way I can tell it to retry on error? If not, is there
> something else I could do to improve the situation?
Date: Wed, 07 Apr 2010 20:41:39 +0200
Message-ID: <8241v3FuddU1_at_mid.individual.net>
On 06.04.2010 16:00, Ian Chard wrote:
> I'm using Oracle 10g, RMAN and TDPO (the TSM client for Oracle). I'm
> trying to do a PITR to a non-production machine, but RMAN is intolerant
> of any media manager problems that crop up, so things like this
>
> ORA-19511: Error received from media manager layer, error text:
> ANS1017E (RC-50) Session rejected: TCP/IP connection failure
>
> and this
>
> ORA-19511: Error received from media manager layer, error text:
> ANS1314E (RC14) File data currently unavailable on server
>
> result in an immediate 'failover to previous backup'. The first error
> was caused by a transient network problem; I suspect the second was just
> bad luck as the file was being reclaimed by the TSM server when TDPO
> asked for it.
>
> Both these errors would have gone away if RMAN had tried again, so is
> there any way I can tell it to retry on error? If not, is there
> something else I could do to improve the situation?
I have worked with a Ora 10g, RMAN and Tivoli on Linux a few years ago. We had so frequent hangups (error message buried somewhere and RMAN just sat there and did nothing) that I created a DB metric to detect that situation. The solution then was to manually kill RMAN. :-( That's of course not a solution to your problem but might be an indication that this kind of integration does not work too well although it sounds great on paper. Does anybody else have experience with that combination?
Kind regards
robert
-- remember.guy do |as, often| as.you_can - without end http://blog.rubybestpractices.com/Received on Wed Apr 07 2010 - 13:41:39 CDT