Last week I was pretty much focusing on setting a lab environment to test SQL 2012 Availability Groups and Failover Cluster Instances.

I had to setup 3 VM’s to play around and out of that one will be a domain controller.

Once all the VM’s were ready and started communicating each other within the domain,I decided to setup the cluster.

The very first process involved in setting up a cluster is to validate the configuration.I went ahead with the validation part.

I was running all these VM’s within my laptop and had no option for a shared storage and I was pretty sure that I can just ignore all the warning which I am going to get for the disks within the cluster validation.

The validation was completed and it came up with 1 expected warning and one additional warning for me.

The additional warning was related to Network

It was simply telling me that I am running on one network card and there is no redundant one available to handle crisis situation.This is a single point of failure.

This is just a warning and will allow you to proceed with setting up the cluster,however its giving you a very important message.

For LAB situations this is normal,however for production deployments you might have to have redundant ones.

                     [Picture courtesy http://support.microsoft.com/kb/813937 ]

Conclusion

Its pretty fine to ignore these warnings when you are doing a LAB setup,however you should be very careful when dealing with highly critical production environments.

There should be a definite plan to implement a cluster upgrade,and its important to have a proper checklist for the same:

1. Take database backup’s including system databases.

2. Script out all the users.

3.Script out all the jobs and other objects if any.

4. Ensure that required .Net framework  and windows installer is installed for both nodes,this will ensure that unexpected downtime is avoided.

4.Start upgrade process from the passive node.

5.Once the passive node is upgraded,failover to the upgraded node and proceed with active node upgrade.

6.Once both nodes are upgraded,ensure that proper failover and failback happens between nodes.

Today I faced a situation where new LUN’s where added to a server and SQL was clustered on top of windows cluster.

I wanted to use the new LUN’s to accommodate database files and I noticed that these LUN’s needs to be added as a dependency for SQL Server Resource.

A possible downtime was on my mind, however in SQL server 2008 cluster the SQL resource will still allow you to add disks as dependency without we taking it offline. This is indeed an awesome feature and it was missing in SQL 2005.

I just love the way SQL Server is improving version – version.