lptech (10/24/2013)
A couple of things caught my eye:1) The jobs would run if the cluster was actually running active/active with the instances running on the opposite nodes.
2) After the failback, the job has to be manually run. This would add an extra thing to keep track of during routine maintenance such as applying security hotfixes.
3) Keeping everything up-to-date on more than a few servers is a pain, and not likely to happen 100% of the time in a medium to large size shop
In an ideal world, these jobs would be scripts that run off of alerts from the monitoring tool. The node and memory information could be kept in a Configuration management Database (CMDB), with blackouts in effect during maintenance windows. Of course, we don't always live in the ideal world.
The job “Alert and Set MaxMemory on Restart” should be scheduled to “Start automatically whenever the SQL Server Agent starts”. That will fire off the jobs to set MaxMem whenever a fail-over occurs, or if you manually move an instance to a node. You should not have to manually run any of the jobs.