SQL Server Performance Forum – Threads Archive

Detailed SQL 2000 Cluster Failover Process

Hello All, I know how to install and configure a SQl 2000 cluster and also do the testing, but does anyone know how actually does the cluster work during the failover process, what process takes place in a sequencial manner…from the failing node to the secondary node. What is going on in SQL, DB locks, transactions, etc….do the transactions failover aswell. Also does the SQL service restart upon failover, if so then it would be like unpluging the power cord on the SQl server and having it reboot….what are the detail steps or procedures that take place. Trying to compare a hardware redundancy versus a failover cluster. Thanks

---

In short you could say that the failover process is the exact same thing as stopping and starting the sql server service on a stand alone server. The same things will happen except that the services then are started on the other node.

---

This long, detailed process, is explained in the book Microsoft Windows 2000 Advanced Server Clustering Services. But essentially, when the SQL Server service fails over, it is like the server was unplugged and then restarted from scratch. There are exceptions to this. In some cases, failover can be more controlled, depending on what caused the failover in the first place. —————————–
Brad M. McGehee, MVP
Webmaster
SQL-Server-Performance.Com

---

Thanks for the explanation, if this is the case (like the server is unplugged) what are the chances of the DB’s getting corrupted. Thanks again…

---

The risk for data corruption is the same as when you unplugg a stand alone server.

---

Generally speaking, the only data you would loose is data that has yet to be comitted to the transaction log, should a failover occur. —————————–
Brad M. McGehee, MVP
Webmaster
SQL-Server-Performance.Com

---

]]>