8.3 Replacing Failed Components
From time to time, it will be necessary to perform hardware maintenance or upgrades on cluster components. Some replacements or upgrades can be performed while the cluster is operative, while others require planned downtime. Make sure you plan all the necessary actions carefully. This will spare you a lot of trouble.
8.3.1 Nodes
When maintaining or upgrading a node, cluster services must usually be stopped on the node. This means down time for the applications usually running on this node, at least during the takeover to other nodes.
Consider the following points when replacing the whole or components of a node:
•Make sure you have at least the same amount of RAM in the replacement system.
•If your applications have been optimized for a particular processor or architecture, ensure that the new node is the same type of system. Uniprocessor applications may run slower on SMP systems.
•Slot capacity of the new node must be the same or better.
•Check the appropriate documentation for a proper adapter placement in your new node.
•The license of your application may be dependent on the CPU ID. You may need to apply for a new license before trying to bring the new node into service.
8.3.2Adapters
In order to replace or add an adapter, the node must be powered off. This means down time for the applications usually running on this node, at least during the takeover to other nodes.
Consider the following points when replacing or adding adapters in a node:
•Make sure that the adapter is your problem and not faulty cabling. Bad cables are much more common than defective adapters. Most network and SSA cables can be changed online. Do some testing, for example, exchange the cables or try to connect to another port in your hub to see if the hub is your problem.