DSSI, MSCP serving and identical nodes



I am trying to workout a transition scenario from my current environment
to a new one that will include my newly adopted vaxes. My goal is to
maintain the cluster up at all times. And I will need to progressively
test and transition hardware from the good old all mighty microvax II to
the not-yet-teenager 4000-600.


I have a few questions:

In a DSSI environment as follows:

ETHERNET +--------------------+----------------------+
| | |
[node1] [node2] [node3]
| |
DSSI +---[R400X-DISKS]----+


node1 has allocation class of 1,
node2 has allocation class of 2,
node3 has allocation class of 3

node1 will be the primary node with node2 probably down most of the
time.

What is the best strategy to decide what allocation class the disks and
tape in the R400X should have ?

What are the implications of them having allocation class 1 versus some
unused number ?

If I have allocation class of 1 for the disks, are there gotchas when
node1 goes down and the serving is done by node2 ?



Also, I would like to be able to bring node2 up so I can bring node1
down for hardware testing/maintenance, but I'd like node3 to still think
it is talking to node1.

Is it possible to bring node2 up with the same SCSNODE , same decnet
address and same IP address and then shutdown node1 as soon as node2 is
up ? If Node2 has its ethernet disconnected while this happens, would it
make it possible ? (I can then just plug node2 and unplug node1 from the ether).

This would happen only during my transition period. Once I am all
installed, node2 will be given a permanent role and its own system root
etc. So not intertested in "is this supported", but rather "how long
can two nodes co-exist with same node-names". If the SCSCSYSTEMID is
different, does it matter ?

Or is this totally impossible and the only way to get around this is to
deal with cluster alias for both decnet and TCPIP ?
.



Relevant Pages

  • 403 Error after extending a SharePoint Site
    ... Create Config DB from NODE1 and connect to the DB in NODE2 ... Test extended site in NODE1 and apply template. ... I've already re-installed SharePoint Portal Server and still get the same ...
    (microsoft.public.sharepoint.windowsservices)
  • Re: Resources not coming online
    ... When I failover from node1 to node2, ... SQL Server, SQL Server Aent and SQL Analysis services do not come online. ... If I failover to node2 again, ... service passwords on node1. ...
    (microsoft.public.sqlserver.clustering)
  • distributed key generation
    ... I have computer network with nodes: Node1, Node2, Node3. ...
    (sci.crypt)
  • Can not ping Node1 to Node2 after failover
    ... Have an NLB with 2 nodes were both nodes are using 1 network card and running ... Cant ping Node1 to Node2 ... Can ping all other devices on the network from both Node1 and Node2 ...
    (microsoft.public.windows.server.clustering)