Dismantle a Windows Cluster and do Manual Failover
Posted on 2009-04-29
I have a client that has two Dell Windows 2003 servers (let's call them SERVER1 and SERVER2) and an MD3000 storage array. Everything is setup in a two node cluster with three disk volumes on the storage array mounted to the active node through the cluster. As this client's needs are changing, we are trying to figure out a way to assign more services to the passive node, but experience has taught me that installing stand alone services to either of a cluster node is a bad idea.
Currently, the recovery time in the event of active node failure is approximately 2 hours, due to limitations of the applications that are running on the servers, they require some manual intervention and configurations on the passive node to redeploy and return to 100% availability. With that in mind, the question came up "What if we remove cluster serivces and just manually re-mount the volumes on the storage array to SERVER2 in the event of failure of SERVER1, and thus freeing up SERVER2 to run some new applications and services while on standby for a SERVER1 failure?" The manual process of redeployment of application on SERVER2 would not take much more time than the current condition.
What is the implications within the Windows operating system to making these changes? Would a re-installation of Windows be required? Is there any other issues that may arise that would steer us away from doing this? Here is the services running in clustered mode:
3 physical disk volumes
- Quorom disk
- Data volume
- SQL volume (SQL server is not part of the cluster)
2 file shares
What would be the best route to accomplish the following:
- Take apart the custer,
- Maintain the Data and SQL volumes on the storage unit
- Re-mount those volumes manually to SERVER1
- Move DHCP server to one single server