Compaq 112726-001 - ProLiant - 6500 Integration Notes page 29

Can data be lost in documents that have not been saved frequently if the server supporting the
volume fails?
Some client applications may fail when the connection to the document is lost. Even
though this data may still be in the client's RAM, data loss could result.
Will the behavior differ for different clients?
Yes. The behavior will differ between clients depending on the type of connection;
stateful vs. stateless. As previously stated, stateful connections must be reestablished.
However, stateless connections, like TCP/IP database clients will seamlessly reconnect to
the clustered service. Both types of connections can use the NetWare Application
Launcher to restart any server-based applications needed by the failed-over clients.
Where can one find info on these products?
http://www.novell.com/products/clusters/
http://www.compaq.com/solutions/enterprise/highavailability.html
What happens to printer jobs when a system running Novell High Availability Server fails?
Any print jobs still in RAM will be lost. The other server will resume all others that have
been written to disk when the volume is mounted on the other active server.
What is the "slunbind" command line parameter on the VINCAIPX line of the
AUTOEXEC.NCF?
With StandbyServer
communication is unidirectional. NHAS uses the same monitoring technology but
requires bi-directional communication because each server in the cluster is monitoring
the other. The slunbind command enables both servers in the cluster to function properly
during a failure.
How do I uninstall NHAS?
The first release of NHAS does not include an automated uninstall. To uninstall NHAS,
the administrator must delete the SYS:\HASERVER directory, then delete the NHAS
command lines from the AUTOEXEC.NCF.
How many times must the Validate NLM be run to validate the cluster?
NHAS must be validated within 30 days of being installed through the Validate NLM and
a Validation Key that is specific to each NetWare serial number. Validate NLM must be
run once on each server, for a total of two times.
0096-0699-A
and
only one server is monitoring the other, so monitoring