Re: VCS and Dataguard

  • From: David Turner <dnt9000@xxxxxxxxx>
  • To: Alex Gorbachev <gorbyx@xxxxxxxxx>
  • Date: Thu, 13 Jul 2006 10:17:24 -0700 (PDT)

I filed a tar with Oracle and believe I have a fix for the problem. However, I 
don't fully understand the problem nor the solution yet. Just that CRS appeared 
to be getting in the way.  I'll send another email when I do. Oracle suggested 
I run
 
 $ORACLE_HOME/bin/localconfig delete
 
 Here's what Oracle wrote:
 
 The minimal version of the CSS (ocssd local version) is installed when the 
software is installed a
 nd is started up using the init 
 scripts so first there is no extra maintainence or monitoring the  customer 
needs to do for this process anyways. This is requi
 red for access to the OCR when having  an ASM configuration or when having a 
CRS
  configuration where this does not remain a "local" configuration as the 
daemon 
 runs out of the CRS home.
 
 You can use the command  "$ORACLE_HOME/bin/localconfig delete" which would 
clear  the 
 local configuration.
 
 
 If anyone understands this better feel free to comment. Dataguard is 
performing switchover nicely. Today I'll attempt to setup the Global Clustering 
Option which allows failover locally and globally.
 
 Dave

----- Original Message ----
From: David Turner <dnt9000@xxxxxxxxx>
To: Alex Gorbachev <gorbyx@xxxxxxxxx>
Cc: oracle-l@xxxxxxxxxxxxx
Sent: Wednesday, July 12, 2006 6:45:38 AM
Subject: Re: VCS and Dataguard

Dataguard now offers third party support for replication such as dataguard. I 
haven't even given VCS control of the database yet. I'm basically just working 
with a couple hosts fronted by a VIP replicating with dataguard to a standby. 
What I tried last night was replacing host=hog with host=hog01. Next, I setup 
dataguard between hog01 and hog03 and last I performed switchover. The 
switchover almost succeeds but the listener dies on the clustered hosts
 
 Here's the listener.log
 
 11-JUL-2006 23:58:30 * service_died * db01 * 12537
 No longer listening on: 
(DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=hog01)(PORT=1521)))
 Listener completed notification to CRS on stop
 
 Why the Cluster Ready Services message in the alert.log? 
 
 Also, could you send me a basic example of a listener.ora without dataguard? I 
just want to be sure I'm not missing something.
 
 Thx, Dave

----- Original Message ----
From: Alex Gorbachev <gorbyx@xxxxxxxxx>
To: dnt9000@xxxxxxxxx
Cc: oracle-l@xxxxxxxxxxxxx
Sent: Wednesday, July 12, 2006 1:10:44 AM
Subject: Re: VCS and Dataguard

From what you described it's not very clear where your problem is.
Dataguard should not have anything to do with VCS unless I there is
something new in 10g. I'm working with 9i standby's with VCS but I
believe that the concept is relevant to 10g as well.

What do you mean by "dataguard tries to find the listener on the
actual hostname". Usually, you should give a connection descriptor for
a listener and this one should point to the listener where your
primary database is registered. So it has nothing to do with VCS.
Could you provide more details on what errors you get, where and when?

2006/7/12, David Turner <dnt9000@xxxxxxxxx>:
> I'm running Oracle 10G with Dataguard and Veritas Cluster Server 4.1 and have 
> run into problems. In the VCS docs I'm told to set the host=<VIP> yet 
> dataguard tries to find the listener on the actual hostname. For example
>
>  vip=hog
>  host1=hog01
>  host2=hog02
>
>  host3=hog03(The standby)
>
>  hog is a virtual ip that points to hog01. I've set the listener to listen on 
> hog and successfully connect with sqlplus from hog03 to db01.hog but can't 
> get dataguard to find the listener. It keeps looking for the listener on 
> hog01 instead of hog.
>
>  Does anyone have a quick howto on VCS and Dataguard? I couldn't find much of 
> anything on metalink and it will be a while before our system guys get 
> through to Veritas.


-- 
Best regards,
Alex Gorbachev

http://blog.oracloid.com



--
//www.freelists.org/webpage/oracle-l





--
//www.freelists.org/webpage/oracle-l


Other related posts: