As much as I've searched, I've not been able to find a single set of best practices for patching production enterprise java systems. I've even heard it described as a black art by one person.
So, my question is, are there defined best practices for providing patches to production systems? What are they? References to any such practices would be gratefully appreciated.
Thanks, Steveo
There is none. It all depends on the kind of your enterprise infrastructure and the quality of service you have to provide. E.g. does the application server provide hot deployments (like OSGi does).
If you have a high availability cluster, you may patch the system node by node at a time, where demand is low. Putting a node down should not be a problem due to controlled failover. Such a process can be automatized. If the demand is unpredictable or high every time, you have to care about extra capacity for the node-updates.
I think, the most common scenario is that you plan for an maintainance window which is communicated throughout all stakeholders. Within the window, the QoS is not guaranteed and all nodes are / the system is updated at once.