|
When defining your virtual SQL servers and monitors in PolyServe why would you NOT use AUTOFAILBACK?
(comments are locked)
|
|
I don't have any experience with PolyServe per se, but generally speaking automatically failing back isn't usally desirable in any cluster. Normally you would want to investigate the cause of the failover and correct any issues that need addressing prior to scheduling a time to fail back. Again, I don't know about PolyServe, but with Windows Server Failover Clustering (or MSCS), a failover requires starting up the instance (running recovery, etc) - while the downtime is minimal, it still necessitates downtime. In the worst case, if your original active node still has an issue and you're employing autofailback you can end up bouncing an instance back and forth between your two nodes - meaning that your HA solution isn't providing very much availability at all. Sounds reasonable enough. I was thinking from a scheduled maintenance point of view where,after bouncing all the physical nodes, all the vSQL instances are back where you want them. But if the nodes are roller-coasting on their own it could be a nightmare.
Oct 29 '09 at 01:52 PM
Blackhawk-17
(comments are locked)
|
