Additional SQL Server features and topics not covered by specific categories
@AmeliaGu-MSFT thank you for your response. I was under impression when you install cluster from scratch, first node has more binaries installed as compared to other nodes. That is why i was hesitant to use this approach. So as per your answer, i dont think there is any dependency like that.
Issue we are having in our current cluster is- Our current application has hardcoded ipaddress and there are lots of unknown areas in the application where we dont know this ipaddress is hardcoded. To workaround this what we do is we have a custom failover vbscript whereby when cluster fails over we assign the same ipaddress to the secondary node's nic card where cluster will failover. This way virtual ipaddress and sql cluster name remains same even after failover. This approach is working fine in our DR environment however not in production.
We see ipaddress assignment is happening correctly on secondary server nic card in PROD however cluster fails to bring sql server resources online. We are thinking there might be some setup issue with the sql cluster install beause of which failover is not happening. Hence we are going to rebuild this cluster the way i explained above.
Steps
- Add 3rd node in wsfc , then add a node in sql server using setup.
- if failover to 3rd node is happening correctly, then remove 1st node from the cluster
- then add another node in wsfc and sql server. if failover is happening correctly to this newer node then remove old second node from the cluster.
is there a better approach to this?