The Question is:
We're running a 4 node cluster (2 alphaserver 100a's, alphaserver 4100, and
DS20e). If one of the four nodes became inoperable for some reason, how would
I re-boot one of the other nodes up as that node? Ex: NODE1, NODE2, NODE3,
NODE4.. If NODE3 failed,
how would I boot up NODE2 as NODE3?
The Answer is :
This sequence depends directly on the hardware configuration.
It may be as simple as selecting another root at the console
boot command, or as involved as a hardware shuffle, an AUTOGEN
pass, and an I/O and layered product reconfiguration. You may
also have to adjust the network connections, depending on how
the paths are configured.
If you wish to rename a node, please see the OpenVMS FAQ for
the areas where the node (host) name can be stored.
Most commonly, the use of the available cluster alias mechanisms
are the prefered approach -- these are transparent and automatic.
The cluster alias approach also allows you to utilize all of the
systems continuously, rather than a comparatively more primitive
hot-standby approach. The cluster alias also helps avoid operator
errors, as having two systems booted from the same root can -- if
the VOTES and EXPECTED_VOTES settings are not entirely correct (and
please see the FAQ) lead to corruptions and untoward behaviour.
The cluster alias approach also avoids having unrecognized failed
hardware, and further allows users to ignore the name of the
particular host servicing the requests.