Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
Jay1Jay
Yak Posting Veteran
50 Posts |
 Posted - 2004-07-26 : 17:02:50
|
| If I would like to install SQL2000 Enterprise edition on my Win2003 Cluster enviorment, is there something I have to do differently. Currently my Win2003 servernames are ServerA and ServerB as the virtual cluster name respectively. Their Physical name are XServerA and XserverB respectively. I have cluster drives D: which is owned by ServerA and Drive F: which is owned by ServerB. I have installed SQL2000Enterprise on XServerA and I saw that installation happend on XServerB at the same time. I thought this was due to Cluster and it might be the case. However, during the install, I specified the Data files to be on Drive D:. So I take it this did an Active -->Passive Install.What if I want an Active-->Active install where some databases are going to be on ServerA and some on ServerB. I'm trying to install SQL2000 now on ServerB and I am unable to run the setup. Any reason why or what should I be looking for. By the way I ran SQLSP3 on ServerA also.Any help is greatly appreciated...Thanks,Jay |
|
|
MuadDBA
628 Posts |
Posted - 2004-07-26 : 17:15:27
|
| You shouldn't have installed the data files to a device that was owned by one server or another, but rather one that was a cluster resource and used by SQL Server.The shared resources are the ones owned by whichever cluster is currently active. This method shouls also allow you to establish an active/active setup should you desire it as well. I am pretty sure with your current setup if the server that owns the D drive dies, you're screwed. |
 |
|
|
Jay1Jay
Yak Posting Veteran
50 Posts |
Posted - 2004-07-26 : 17:21:23
|
D: drive is a shared cluster resource drive, so say if my ServerA goes down the D: drive where the data files are fails over automatically to ServerB. Question I have is since I did the install only on ServerA, how did cluster get install on my ServerB. Is it because of cluster. Also, what if I want to install SQL on ServerB now, will that be considered Instance on ServerB since SQL is already present on the server from the ServerA install. I'm just confused, I know how cluster works and that is setup correctly. But confused as far as SQL install works and instances,etc... So any help is appreciated..quote:
Originally posted by crazyjoe
You shouldn't have installed the data files to a device that was owned by one server or another, but rather one that was a cluster resource and used by SQL Server.The shared resources are the ones owned by whichever cluster is currently active. This method shouls also allow you to establish an active/active setup should you desire it as well. I am pretty sure with your current setup if the server that owns the D drive dies, you're screwed.
|
|
|
|
eyechart
Master Smack Fu Yak Hacker
3575 Posts |
Posted - 2004-07-26 : 17:48:10
|
| I just wanted to take a step backward and have you check a couple of things.In the cluster administrator, you should have one group called 'Cluster Group', this is the default name and should have 3 resources in it:1. Cluster IP Address2. Cluster name3. Disk Q (or whatever driver letter you gave the Qurorum disk)Keep in mind that this cluster IP address and Cluster name have nothing to do with the Virtual IP address and Virtual name you will be giving your SQL Server instance when you install it.A typical SQL server (active/passive config) has 4 IP address/names minimum.1. Node A (physical address/hostname)2. Node B (physical address/hostname)3. Cluster VIP/name (virtual address/name in "Cluster Group")4. SQL VIP/name (virtual address/name for SQL instance)btw, Quorum disk should probably be drive Q, that is somewhat of a standard these days. It also doesn't need to be very large, a 400MB quorum would be about 5x what is really needed.If you have a 2nd cluster group (sounds like you do) this is where you will be installing your first SQL Server instance. make sure that this cluster group has atleast one drive associated with it. Make sure it is of an appropriate size to hold your shared resources (datafiles, logfiles, etc.). You will also need to have determined what your virtual ip for this instance will be, along with a virtual name.If this is what you have done, then you are finished. This is an active/passive config because you only have one instance installed. Installing a 2nd instance would allow you to run an active/active configuration. Installing a 2nd instance means you need to come up with a new shared disk resource. This cannot be drive D: or drive F: in your current config. Each instance of SQL server in a cluster has at a minimum these 4 resources:1. unique virtual IP address2. unique virtual hostname3. shared disk resouce not used by any other cluster group4. One instance of SQL Server installedkeep in mind that even when you run active/active you will not be pointed to the same database from two separate nodes. MSCS uses a shared nothing approach to clustering, active/active simply means that both sides of the cluster are active hosting separate physical databases.You might want to go to the microsoft website and do some searches for "sql server mscs" and read the articles there. You can start with these:http://msdn.microsoft.com/library/en-us/adminsql/ad_clustering_7t9v.asphttp://www.microsoft.com/technet/prodtechnol/sql/2000/maintain/failclus.mspx-ec |
|
|
|
Jay1Jay
Yak Posting Veteran
50 Posts |
Posted - 2004-07-27 : 10:31:04
|
Thanks for the details.... Let me explain you my sceniro and you tell me what you think. I have 2 servers with cluster installed. Server A has a shared resource D: drive which is owned by ServerA but on external sub-system. Server B has a shared resource F: drive which is owned by Server B and also on the the same external sub-system. Both of these drives are owned by its owner with cluster. So therefore I have 3 different cluster group. First group(NODE) which includes the virtual name for Server A, Virtual IP and the D: drive. Second group(node) have virtual name for Server B, virtual IP and F: drive. 3rd group has the Quoruom drive Q:, for clustering. I installed SQL 2000 Enterprise on ServerA my first node, and gave it its virtual SQL name and IP, used D: drive as my data drive. Also installed SP3. When I did this, I saw that SQL got installed on Server B as well. I tested failover and failback and it tested successfully. However, when I went to install SQL on Server B, it only prompted me for an instance name instead of using default. So it is true to say that after the first install of SQL on either node, if I install SQL it is considered as an instance. Thanks for your help....Jayquote:
Originally posted by eyechart
I just wanted to take a step backward and have you check a couple of things.In the cluster administrator, you should have one group called 'Cluster Group', this is the default name and should have 3 resources in it:1. Cluster IP Address2. Cluster name3. Disk Q (or whatever driver letter you gave the Qurorum disk)Keep in mind that this cluster IP address and Cluster name have nothing to do with the Virtual IP address and Virtual name you will be giving your SQL Server instance when you install it.A typical SQL server (active/passive config) has 4 IP address/names minimum.1. Node A (physical address/hostname)2. Node B (physical address/hostname)3. Cluster VIP/name (virtual address/name in "Cluster Group")4. SQL VIP/name (virtual address/name for SQL instance)btw, Quorum disk should probably be drive Q, that is somewhat of a standard these days. It also doesn't need to be very large, a 400MB quorum would be about 5x what is really needed.If you have a 2nd cluster group (sounds like you do) this is where you will be installing your first SQL Server instance. make sure that this cluster group has atleast one drive associated with it. Make sure it is of an appropriate size to hold your shared resources (datafiles, logfiles, etc.). You will also need to have determined what your virtual ip for this instance will be, along with a virtual name.If this is what you have done, then you are finished. This is an active/passive config because you only have one instance installed. Installing a 2nd instance would allow you to run an active/active configuration. Installing a 2nd instance means you need to come up with a new shared disk resource. This cannot be drive D: or drive F: in your current config. Each instance of SQL server in a cluster has at a minimum these 4 resources:1. unique virtual IP address2. unique virtual hostname3. shared disk resouce not used by any other cluster group4. One instance of SQL Server installedkeep in mind that even when you run active/active you will not be pointed to the same database from two separate nodes. MSCS uses a shared nothing approach to clustering, active/active simply means that both sides of the cluster are active hosting separate physical databases.You might want to go to the microsoft website and do some searches for "sql server mscs" and read the articles there. You can start with these:http://msdn.microsoft.com/library/en-us/adminsql/ad_clustering_7t9v.asphttp://www.microsoft.com/technet/prodtechnol/sql/2000/maintain/failclus.mspx-ec
|
|
|
|
|
|
|
|
|
SQL Cluster Installl.....