Re: Issues creating a new failover cluster on the same server
- From: Dan <Dan@xxxxxxxxxxxxxxxxxxxxxxxxx>
- Date: Mon, 12 Feb 2007 12:27:01 -0800
Thanks for all your input so far and I apologize that I have my terminology
wrong. After reading through the documentation and what you have posted let
me try and repost what our current setup is.
We have two nodes in our active passive cluster, sq04a and sq04p, sq04a is
the active node and sq04p is the passive node.
We created a cluster called CHSSQ04 and inside of this cluster we created
several groups; CHSSQ04, Cluster, and MSDTC.
The group Cluster has our Quorum Disk, Cluster IP, and Cluster Name. The
group called MSDTC has our MSDTC disk, MSDTC IP Address, MSDTC Network name,
and MSDTC Resource.
Note: All of the disks are also part of the sans, and the sans has been set
to allow the two nodes to see the disks, but only the node that is active can
be using the disks.
For each of the groups we assigned the two nodes as the preferred owner.
In our sans we created 5 more drives (LUNS) and give both nodes permission
to use the drives, just like the other ones listed above. So when we open up
disk management on both nodes we see the drives, but only the active server
can see drive name and if its been formatted.
We then went to the active node and formatted the drives and assigned drive
letters. The passive node at this point still only sees the drives as
unformatted and with no drive letters assigned. We then created disk
resources for the 5 disks and assigned them to the CHSQ04 group, since this
is where the SQL instance is going to be at.
Note: yes we do have two things in our setup named CHSSQ04, the actual
cluster is called CHSSQ04 and then there is a group called chssq04 group.
By doing so the SQL installer created an instance and setup the SQL IP
Address, SQL Network Name, SQL Server (with the instance name CHSNextgen),
SQL Server Agent (CHSNextgen), and SQL Server Fulltext (CHSNextgen).
So now we have a SQL server with the name vsSQL04 with a public ip address
and an instance called CHSNextgen. We then setup our database and move are
data into and set the clients to use vsSQL04\chsnextgen to connect to the db.
The SQL installer also created binaries for the instance on the c:\ drive of
both nodes. It looks like this;
C:\Program Files\Microsoft SQL Server>dir
Volume in drive C has no label.
Volume Serial Number is 44E8-2F28
Directory of C:\Program Files\Microsoft SQL Server
02/06/2007 10:18 AM <DIR> .
02/06/2007 10:18 AM <DIR> ..
11/10/2006 10:22 AM <DIR> 80
01/16/2007 10:25 PM <DIR> 90
11/10/2006 10:26 AM <DIR> MSSQL.1
11/20/2006 08:49 AM 4,286,976 SQLUASetup.msi
1 File(s) 4,286,976 bytes
6 Dir(s) 109,489,816,576 bytes free
And on G: we have the following,
G:\Microsoft SQL Server\MSSQL.1\MSSQL>dir
Volume in drive G is SQL Data Primary
Volume Serial Number is FE52-BCF2
Directory of G:\Microsoft SQL Server\MSSQL.1\MSSQL
02/04/2007 09:59 AM <DIR> .
02/04/2007 09:59 AM <DIR> ..
02/05/2007 12:58 AM <DIR> Backup
02/04/2007 09:59 AM <DIR> Data
11/10/2006 10:26 AM <DIR> FTData
11/10/2006 10:26 AM <DIR> JOBS
02/12/2007 02:07 PM <DIR> LOG
11/10/2006 10:26 AM <DIR> repldata
0 File(s) 0 bytes
8 Dir(s) 28,332,949,504 bytes free
This works great and fails over the way it should. In our test we turned off
sq04a and the cluster failed over to sq04p with all the drive letters and
instance. So everything worked the way it should.
What we are trying to do is create another SQL fail over instance on the
same active node so that when someone is in the SQL management console for
the new instance they can not see the drives nor the data in the first SQL
instance named vsSQL04. What I did to try and accomplish this was create a
new group in the cluster called chnsq04, then assign more drives to this
group for the new instance. I started the setup and selected SQL fail over
cluster, create a new instance, and then selected the new chnsq04 group. The
setup created the instance and placed the files and folders where they need
to be, but after the install finished the SQL browser stopped and caused our
first instance in the other group to stop accepting and dropped connections.
I have since removed the new instance and group and corrected a few issues
with the admin port and the SQL browser
Now you guys have both noted that there should be a SQL browser for each
instance, well I only have one. If I look at the SQL server browser
properties from the SQL server configuration manager, I notice that the dump
directory is set to use G:\Microsoft SQL Server\MSSQL.1\MSSQL\LOG\. When I
had installed the second instance it changed this value to H:\Microsoft SQL
Server\MSSQL.2\MSSQL\LOG\ which caused my first instance to die. Also my
adminConnection/tcp port was deleted, which I have already resolved.
On a side note, our setup (before we attempted to create another fail over
instance) was done by EMC/Dell according to a MS guide. The only thing I was
trying to do was mimic the same setup for a new group so I could have a new
virtual server name and instance, if this is at all possible.
I hope this clarifies what is going on and what I am trying to accomplish.
.
- References:
- Issues creating a new failover cluster on the same server
- From: Dan
- Re: Issues creating a new failover cluster on the same server
- From: Anthony Thomas
- Re: Issues creating a new failover cluster on the same server
- From: Dan
- Re: Issues creating a new failover cluster on the same server
- From: Anthony Thomas
- Issues creating a new failover cluster on the same server
- Prev by Date: Re: SQL Server 2005 Cluster requirements
- Next by Date: for hazel: really delicious premium nntp server access - icgu itfo - (1/1)
- Previous by thread: Re: Issues creating a new failover cluster on the same server
- Next by thread: Re: MSDTC / sql2000 active/active
- Index(es):
Relevant Pages
|
