Cluster related queries can be challenging to troubleshoot, and it’s not uncommon to receive inquiries about them. Recently, I received an email from a friend who was experiencing an issue with their SQL Server 2008 R2 SP1 Cluster. During failover from NodeA to NodeB or vice versa, the system databases were going into upgrade mode, resulting in a login failure for the ‘sa’ user. In the email, my friend asked for help in understanding why this was happening and how to prevent it from occurring with every failover.

After analyzing the error, I asked my friend to share the ERRORLOG files to investigate further. The ERRORLOG files contain valuable information about the state of the SQL Server instance. Once I received the files, I noticed something interesting. The versions of SQL Server on the two nodes were different. NodeA had version 10.50.2500.0 (X64), while NodeB had version 10.50.1600.1 (X64).

This discrepancy in versions can occur if all nodes in the cluster are not patched up to the same version of SQL Server. To resolve this issue, it is necessary to apply the same service pack or patch on all nodes in the cluster. This ensures that the version reported in the ERRORLOG files is consistent across all nodes.

In my friend’s case, I recommended applying SP1 on NodeB (SQLCRM-N03), which brought it to the same version as NodeA (SQLCRM-N02). After applying the patch, the error disappeared, and the failover process no longer triggered the upgrade mode for the system databases.

If you have encountered a similar situation, it is crucial to ensure that all nodes in your SQL Server cluster are running the same version. Inconsistencies in versions can lead to unexpected behavior during failover and cause issues like the one described above.

When troubleshooting cluster-related queries, it is always helpful to check the ERRORLOG files for any relevant information. These files can provide valuable insights into the state of your SQL Server instance and help identify the root cause of any issues.

Have you ever faced a similar situation? How did you troubleshoot it? I would love to hear about your experiences and any alternative mechanisms you used to solve the problem. Feel free to share your thoughts and insights in the comments below.