RE: How to install 2 set of Oracle Clusterware into the same machines?
Date: Wed, 30 Mar 2011 08:23:38 +0100
Message-ID: <000001cbeeab$694f63e0$3bee2ba0$_at_the-playground.de>
Ouch, that hurts purely from listening to it. Please don't get me wrong, I don't want to bash you or your setup in any way, it's just the written equivalent of showing some real astonishment.
I actually had to twist peoples' arms at my current site to avoid being pressed into a similar situation where storage tech wanted to replicate _all_ SAN storage, including OCR/voting disks and Oracle binaries (but that wasn't across versions). I managed to convince them eventually that replication of DATA is ok, but not METADATA. We also had our own set of Oracle binaries in the end-rolling upgrade and "failback policy" in case of one-off patches seriously brake things on the primary.
Thinking clearly about the invocation of DR in your setup I have some doubts if it can be pulled off.
First, the OCR stores lots of references to the host names so you'd have to have your standby servers use the same as production. Which implies that can't really be "active" but should sit quietly in a corner. Secondly, you need to be careful with name resolution when using the standby hosts as "primary" - your TTL has to be pretty low And then, how would you stop the active clusterware stack and invoke the "DR" one? There are lots of files (named pipes and the like) that are in a default location, and I don't think Clusterware has been developed to coexist with another version of itself, especially in 10.2.
I'd be very curious if that works, and how you are planning on invoking the other Clusterware stack.
Best regards,
Martin
> -----Original Message-----
> From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org]
> On Behalf Of Leyi Zhang (Kamus)
> Sent: 30 March 2011 07:42
> To: Yechiel Adar
> Cc: oracle_l
> Subject: Re: How to install 2 set of Oracle Clusterware into the same
> machines?
>
> Sorry for confuse you guys, I have to make it more clear.
>
> the situation is complicated. I didn't mention the whole story in my
original
> post because I think it is not related to technology.
>
> Actually the story is:
> these 2 machines are the backup site of a 2-node 10.2.0.4 RAC product
> database, using the storage replication method instead of using dataguard,
> the OCR and Voting and all the datafiles are synced in disk volume level,
the
> disk volume cannot be opened, and cannot be changed when they are in
> sync mode. In case the product database is crash and can't recover easily,
we
> will open the synced volume, startup the clusterware and the database. We
> don't want to lose this backup site, so we can't deinstall the 10.2.0.4
> clusterware.
>
> And now we want to use these 2 machines to build a new 10.2.0.5 RAC, so
> we have to install another clusterware and another RAC. since the synced
> storage can't be opened, so I assume it's just like the storage is not
> connected.
>
> --
> Kamus <kamusis_at_gmail.com>
>
> Visit my blog for more : http://www.dbform.com Join ACOUG:
> http://www.acoug.org
>
>
>
> On Wed, Mar 30, 2011 at 1:44 PM, Yechiel Adar <adar666_at_inter.net.il>
> wrote:
> > I really do not get it.
> > Why do you want to use 10.2.0.4 after you install 10.2.0.5?
> > Just define the resources in the 10.2.0.5 and stay with this.
> > Why do you want to fall back to a previous version?
> >
> > Yechiel Adar
> > Israel
> >
> >
> > On 29/03/2011 15:23, Leyi Zhang (Kamus) wrote:
> >>
> >> Hi experts
> >>
> >> For some reason, we have to install another Oracle clusterware into
> >> the existing cluster.
> >>
> >> 1. we have 2 machines have installed oracle clusterware 10.2.0.4, but
> >> this clusterware is not running, the storage which contains OCR and
> >> Voting is not connected, but sometime we will connect it again, so we
> >> can't deinstall this clusterware.
> >> 2. we decide to use these 2 machines, we connect a new storage, and
> >> plan to install a new 10.2.0.5 clusterware using a new user account.
> >> 3. If sometime the old storage is back, we plan to shutdown the new
> >> clusterware stack and change some script and open the old clusterware.
> >>
> >> Is it possible to implement? I know we can't running 2 oracle
> >> clusterware in the same cluster at the same time, but if one
> >> clusterware is not running, how about install and run the second one?
> >> If is possible, How many scripts or files need to be backuped to
> >> ensure the old clusterware not broken? like /etc/init.d/init.crs,
> >> init.crsd, init.cssd, init.evmd ?
> >>
> >> Any suggestion will be very appreciated.
> >>
> >> --
> >> Kamus<kamusis_at_gmail.com>
> >>
> >> Visit my blog for more : http://www.dbform.com Join ACOUG:
> >> http://www.acoug.org
> >> --
> >> http://www.freelists.org/webpage/oracle-l
> >>
> >>
> >>
> >
> --
> http://www.freelists.org/webpage/oracle-l
>
-- http://www.freelists.org/webpage/oracle-lReceived on Wed Mar 30 2011 - 02:23:38 CDT