[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference decwet::winnt-clusters

Title:WinNT-Clusters
Notice:Info directories moved to DECWET::SHARE1$:[NT_CLSTR]
Moderator:DECWET::CAPPELLOF
Created:Thu Oct 19 1995
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:863
Total number of notes:3478

741.0. "Oracle 7.2 installation checklist - pointer?" by MIPSBX::"Nilssons@mail.dec.com" (Sven Nilsson) Mon Apr 14 1997 20:51

Hello.

I have problem to verify an Oracle Workgroupserver 7.2
installation on 3.51 SP5 1.0 SP2 Alpha cluster running
HSZ40 dual controller.

Customer reports that the databaseserver starts before the disk is ready.

I have not found any good Oracle installation manuals and the NT Cluster
manuals starts when the database server is already installed with creating 
failover groups and so on..

Is there an installation checklist, pointer to good manual somewhere?
Which services to expect - how are they started manual/automatic?
What part of Oracle on common cluster disk and what on local disk?
Configurations files, what to expect - where? contents?

questions, questions questions...

Lots of Thanks in advance!

/Sven Nilsson in the ceNTer of Sweden

(I have read note 479 - good stuff, but what to check before that checklist?)
[Posted by WWW Notes gateway]
T.RTitleUserPersonal
Name
DateLines
741.1Any detailed tips...DEKVC::MANKIDOMon Apr 21 1997 11:2715
    Hello,
    
    Like .0, I'm installing the WNT cluster V1.1 with WNT 4.0 and
    Oracle Enterprise V7.3. To setup this, I followed up ch3 and 
    ch9 in Admin's Guide issued at Februray 1997 with Oracle Engineer.
    
    But the shared Oracle DB can't be opened from primary server.
    
    So, Is there any detailed tips for installation and Admin for
    Oracle failover on above version?
    
    Regards,
    
    MK DO                              
    
741.2More inst questionsNNTPD::"Nilssons@mail.dec.com"Sven NilssonTue Apr 22 1997 20:52119
Hello again

[NT Cluster 1.0]

I still have some questions before I know how to install
Oracle and make some successful failovers..

I know that the Customer is expected to support with Oracle
DBA expertise but I would like to know a little bit more..

Questions..

During installation of Oracle Workgroup Server 7.2
one of the first questions after entering the passwords
Oracle will use you get the question

Host:?

Should I enter the nodename of the Server I'm installing on
or the ClusterAlias?

Question 2..
If I install Oracle Server on one of the nodes giving
Host: Nodename during inst I get successful installation
and an database instance alias

wgs_nodename_ORCL

eg wgs_NodeA_ORCL on node Node A
     wgs_NodeB_ORCL on node NodeB

Should one keep these sample databases/instances
on local drive and create a NEW instance on the
cluster disk or should I somehow move the ORCL instance to
clusterdrive?

(The examples in manual always talkes about the
Oracle sid: orcl and Pipe Name: ORAPIPE which make
me confused, if it is the instance created during
installation that should be the one failing over..)

Question 3.

The Services started on each node

sid ORCL instance on local drive...
OracleServiceORCL    automatic startup
OracleTNSListener      automatic startup
OracleStartORCL         automatic startup

sid NTCL instance on cluster drive
OracleServiceNTCL    automatic startup
OracleStartNTCL         MANUAL startup

The OracleStartNTCL being the service started via the Oracle
Failover dll, right?

Question 4
Running NT Cluster 1.0 SP2 I get the impression
only Named Pipe is supported?
Using Protocol TCP have to wait until 1.1 and IPfailover, right?

Question 5.. bear with me..:-)
"To access a server from a client, you need to configure a database alias
useing the SQL*NET Easy Configuraton tool and you need to modify the
listener.ora file in \orant\network\admin directory.."

On the SERVER, right?
e g
(ADDRESS=
	  (COMMUNITY= NMP.world)
	  (PROTOCOL= NMP)
          (SERVER= ying)          <----ServerName? 
          (PIPE= ORAPIPE)

Does it mean ..SERVER=yang) on other Cluster node?

Question 6..
TNSnames.ORA file
about the same questions as question no 5..

e g
toto_np.world = 
  (DESCRIPTION = 
    (ADDRESS_LIST = 
        (ADDRESS = 
          (COMMUNITY = nmp.world)
          (PROTOCOL = NMP)
          (Server = YING)     <--- ServerName? ClusterName?
          (Pipe = ORAPIPE)
        )
    )
    (CONNECT_DATA = (SID = TOTO)
    )
  )
toto.world = 
  (DESCRIPTION = 
    (ADDRESS_LIST = 
        (ADDRESS = 
          (COMMUNITY = nmp.world)
          (PROTOCOL = NMP)
          (Server = ying)
          (Pipe = ORAPIPE)
        )
    )
    (CONNECT_DATA = (SID = toto)
    )
  )


toto.world and toto_np.world what do the 2 very simular
entries do? Both working on sid TOTO, right?
Both using Named Pipe, ORAPIPE and both on server YING..
comments pls?

Of course a lot of thanks in advance!

Regards/Sven 
[Posted by WWW Notes gateway]
741.3some helpBLUFSH::MCCROREYFri May 02 1997 13:3329
I'll try to help a little but I'm working from memory as I do not have
an Oracle installation in fornt of me. First a few suggestions:
1. Get the latest NT Clusters documentation and read it carefully. There
   is good information there that explains the Listener.ora file and the
   tnsnames.ora file. I believe that it also points out information about
   host name verse cluster alias name.
2. Install Oracle on each cluster member as if they were separate systems.
   Get oracle working on each node first before you begin to think about
   clusters and failover.
3. You can use one of the databases created during installation or create a
   new one to be used for failover. In either case, the database must of
   course be on a shared disk. I don't remember how to move a database but
   I know you can do it. The problem is more than copying files. I think the
   Oracle documentation explains how. 
4. You must create a listener.ora entry for this shared database. The entry
   should be identical on both nodes. It's stored in the Oracle home directory
   on each system. You also create a failover group with the shared disk and
   Oracle failover object. See the clusters doc for details. (The listener.ora
   file is where you use the cluster alias name)
5. Set up a matching entry in the tnsnames.ora file for the clients that will
   use the database. (Again, it's in the clusters doc)

Some other answers:
- The sid "ORCL" and pipename "ORAPIPE" are default values that Oracle uses.
  You need to use other values to have more than one database on a system.
- Yes, only clusters 1.1 supports IP failover.

Good luck,
 - Jack
741.4One more tipTAVEIS::SHALOMSun May 18 1997 09:4014
Hi,

	HAving fought the battle , lose it three times and finally win it, I
can only offer one additional advice:

	BE VERY CAREFULL WITH upper/LOWER case leeters in the sid names. Do not
use CLST Clst, ClST and cLsT interchangably.

	This single tip caused most of the grief, and adhering toit - after
reading the docs etc. - did the trick.

	Good luck - it is much easier than SQL...

Shalom