Have you came across this issue ?
in case the Ethernet management port of one of the node is not available (due to network or other issue), it will cause malfunction of the cluster dynamic DNS entry, until the node will be removed from the cluster.
Your observation is right; there has been a discussion at the
Isilion Technical Users *Google* Group in April this year,
under the subject "Isilon-Users Is there an isi command to move
the SmartConnect Service IP to another cluster node?"
You can't easily move the SmartConnect DNS service around in the cluster.
Some additional thoughts:
If you can make a diskless node (i.e. accelerator) the node 1,
then you can "failover" the SmartConnect service to node 2
by stopping node 1, WITHOUT having to re-protect and later
re-balance the cluster.
One could also think of preparing a separate DNS service
for the SmartConnect zone(s), but outside the Isilon,
and of course with static (or round robin) addresses
rather than load-balancing. In the main DNS server
one would add this backup server as second provider for the zone(s).
At least the client will be able to do lookups and to connect,
until the cluster's SmartConnect DNS become available again.
thank you Peter,
still why should a network outage of a port cause SC to be stuck,
i had a similar issue when adding new nodes node 1 froze not releasing some routing params and it had to be halted.
Also the NAS services do not failover in case of a mere network interruption,
only in case of a total outage of a node.
But from a user's point of view I fully agree with you.
For example, it shouldn't be too hard running three instances
of SC on the cluster and have all of them registered with
the main DNS server.
How about taking this a feature request / improvement to our sales reps?