Re: Failover using cloning and virtual host

  • From: "Jurijs Velikanovs" <j.velikanovs@xxxxxxxxx>
  • To: ora-apps-dba@xxxxxxxxxxxxx
  • Date: Tue, 26 Jun 2007 09:56:47 +0300

Hananto,

Just to confirm that you are saying
Except ICM, you can set CMs node that activated to virtual hostname in that
That means that you can't set virtual node as primary/secondary for ICM?
How you would let ICM to run on virtual host then, or it isn't possible ?

Jurijs


On 6/25/07, Hananto Wicaksono <hananto.wicaksono@xxxxxxxxx> wrote:
Hi Joseph,

Except ICM, you can set CMs node that activated to virtual hostname in that
form. If you keep it null, then administer form will show you it run on your
current persistent node name automatically. So that's why it is differrent.

Don't forget to restart concurrent managers if you make changes to this one
(if neccessarry all apps service), so your change will take effect.

Thanks.

cheers,

Hananto.


On 6/26/07, Joseph Amalraj <joseph@xxxxxxxxxxxxxx> wrote:
>
>
>
>
> Hananto,
>
>
>
> Thanks.
>
>
>
> In this screen, only the following are running on the physical host. For
others, primary_node_name is null or virtual host name.
>
>
>
> Internal Monitor: physical_host_name
>
> Service Manager: physical_host_name
>
>
>
>
>
> This screen show a different output compared to "Administer" screen.
>
>
>
> Joseph
>
>
>
> ________________________________

>
> From: ora-apps-dba-bounce@xxxxxxxxxxxxx
[mailto:ora-apps-dba-bounce@xxxxxxxxxxxxx] On Behalf Of
Hananto Wicaksono
> Sent: Monday, June 25, 2007 1:14 PM
>
> To: ora-apps-dba@xxxxxxxxxxxxx
> Subject: Re: Failover using cloning and virtual host
>
> To: ora-apps-dba@xxxxxxxxxxxxx
> Subject: Re: Failover using cloning and virtual host
>
>
>
>
>
>
>
>
>
> Hi Joseph,
>
>
>
>
>
> as my previous updates, you can change using concurrent > manager > define
on apps level.
>
>
> Change them one by one to virtual host since the standby server will also
using virtual host when failover occurred.
>
>
>
>
>
> Since you are using single node, you don't need metric server and metric
client.
>
>
>
> Cheers,
>
>
>
>
>
> Hananto.
>
>
>
> On 6/26/07, Joseph Amalraj < joseph@xxxxxxxxxxxxxx> wrote:
>
>
>
> Thanks; that was very informative.
>
>
>
> In the current configuration
>
>
>
> Following concurrent managers are running on the virtual node
>
> Apache Listener, Service Manager, Forms Listener, Metric Client, Metrics
Server, Report Server
>
>
>
> The rest of the managers are running on the physical node.
>
>
>
> How do I change the node for the concurrent managers other than ICM to
virtual node? Can it be changed after failover?
>
>
>
> Thanks
>
>
>
> Joseph
>
> ________________________________

>
> From: ora-apps-dba-bounce@xxxxxxxxxxxxx
[mailto:ora-apps-dba-bounce@xxxxxxxxxxxxx] On Behalf Of
Hananto Wicaksono
> Sent: Monday, June 25, 2007 11:33 AM
>
>
>
> To: ora-apps-dba@xxxxxxxxxxxxx
> Subject: Re: Failover using cloning and virtual host
>
>
>
>
>
> Hi Joseph,
>
>
>
>
>
> You can see on Apps :
>
>
> System administrator resp. > Concurrent > Manager > Administer and see at
node column. You can configure all CMs at Concurrent > Manager > Define.
>
>
>
>
>
> Cheers,
>
>
>
>
>
> Hananto.
>
>
>
>
>
>
> On 6/25/07, Jurijs Velikanovs < j.velikanovs@xxxxxxxxx> wrote:
>
> Hi,
>
> Honestly speaking I didn't setup Apps processes in Cold Cluster
> configuration (Virtual Host) so my comments in that case are purely
> theoretical. It sounds like Hananto have more experience.
>
> HA for Apps parts in most of my projects has been ensured by backup &
> recovery strategy for really small systems and it doesn't make sense
> to make more expensive solutions in that case or using Load Balances
> (Web, Forms) & PCP (Concurrent Processing) for enterprise systems.
>
> Cold Clusters however is kind of trade off between to mentioned in above.
>
> Jurijs
>
> On 6/25/07, Hananto Wicaksono <hananto.wicaksono@xxxxxxxxx> wrote:
> > Hi Jurijs,
> >
> > Thanks for the information. That is just in case if some patches or
maybe
> > autoconfig can overwrite the TNS on Apps Tier, but I am also not sure. I
> > think it is good also to put our customization on IFILE. :)
> >
> > Joseph,
> > From my experience, when configure the failover as described, make sure
that
> > ICM is running using physical node name and other concurrent managers
are
> > running on virtual node name. ICM cannot run when using virtual
hostname, so
> > leave it since it will be automatically detect on which node name it is
> > actually running. But for other concurrent managers must refer to
virtual
> > hostname so it can failover. As Jurijs said, check also the FND_NODES.
It
> > must contains all the physical nodes name as well as virtual hostname.
> >
> > CMIIW, you are not using PCP, right ? Just running single node but with
> > failover capability to standby server (active-pasive). And Yes, always
refer
> > to the virtual hostname when setting the xml file. None of the parameter
> > refer to the persistent IP/hostname (physical).
> >
> > Cheers,
> >
> > Hananto.
> >
> > On 6/25/07, Joseph Amalraj < joseph@xxxxxxxxxxxxxx> wrote:
> > > The CM host in the <SID>_hostname.xml file is the name of the Virtual
> > host.
> > >
> > > Joseph
> > >
> > > -----Original Message-----
> > > From: ora-apps-dba-bounce@xxxxxxxxxxxxx
> > > [mailto:ora-apps-dba-bounce@xxxxxxxxxxxxx] On Behalf
Of
> > Jurijs Velikanovs
> > > Sent: Monday, June 25, 2007 6:59 AM
> > > To: ora-apps-dba@xxxxxxxxxxxxx
> > > Subject: Re: Failover using cloning and virtual host
> > >
> > > Joseph,
> > >
> > > It sounds like your Apps configuration is not 100% running on Virtual
> > host.
> > > Your CM-s are running on physical host, is it right?
> > >
> > > If you might want to play with Parallel Concurrent Processing feature
> > > which allow you to run CM-s on several hosts at the same time.
> > > Or if you would like to keep those running on physical host then just
> > > before running postclone actions you need to execute
> > > fnd_conc_clone.setup_clean; procedure in order to clean old records in
> > > FND_NODES table. Postclone calls autoconfig which adds proper hostname
> > > to that table.
> > >
> > > If you CM-s have been interrupted in abort mode (shutting down a node
> > > for example) then you might considering to run cmclean.sql (available
> > > on Mentalink just enter script name in the search field) script just
> > > before starting up CM-s in the target node.
> > >
> > > Jurijs
> > >
> > >
> > > On 6/22/07, Joseph Amalraj < joseph@xxxxxxxxxxxxxx> wrote:
> > > >
> > > > Hi,
> > > >
> > > > We have a EBS single node environment. This uses same virtual node
for
> > all
> > > > services. A failover environment is currently configured so that the
> > > virtual
> > > > node (and its ip address) fails over to the standby site. We ran
> > preclone
> > > > and rapid clone for the failover. All services came up except for
> > > concurrent
> > > > manager processes. I guess I should have registered the physical
host
> > name
> > > > through OAM as a post clone step.
> > > >
> > > > Please give me your input how this should be configured.
> > > >
> > > >
> > > > Thanks
> > > >
> > > > Joseph Amalraj
> > > >
> > > >
> > > >
> > >
> > >
> > > --
> > > Yury
> > > +371 29268222 (+2 GMT)
> > > ============================================
> > > http://otn.oracle.com/ocm/jvelikanovs.html
> > >
> > >
> > >
> > >
> >
> >
>
>
> --
> Yury
> +371 29268222 (+2 GMT)
> ============================================
> http://otn.oracle.com/ocm/jvelikanovs.html
>
>
>
>




--
Yury
+371 29268222 (+2 GMT)
============================================
http://otn.oracle.com/ocm/jvelikanovs.html

Other related posts: