HP StorageWorks Storage Mirroring application notes Guidelines for networking and failover (T2558-96063, February 2008)

12
documentation on how to script changes to DNS records other than the Windows 2000 DDNS solution.
However, since DNS zone files are text-based, they can be manipulated with any scripting language that
can open, parse, and write to a text file.
Networking failover using the Storage Mirroring online recovery features
In order to ensure a successful failover/failback using the Storage Mirroring online recovery features over
a WAN, you should verify the following:
• You must have a public and private network configuration.
• The private configuration must be in a different subnet so that clients cannot access it.
• Verify that you can ping the source from the target over both the public and private IP addresses.
• Verify that you can ping the target from the source over both the public and private IP addresses.
• Verify that you can resolve by name as well as by IP address.
• Set the private NIC to be the default NIC used by Storage Mirroring. This ensures that failover, failback,
and restore occur on only the private connection.
• In the monitor settings, verify that you have
not selected the IP address to failover across the WAN.
NOTE: Selecting not to failover the IP address requires you to use DNS scripting. For more information,
see Failover and DNS on page 7
• Set the failover monitor to monitor the public address only.
• Set the failover and failback and restore to occur on only the private connection.
• On the source, modify the
SourceNetworkRefresh.bat script to work with the public NIC name of
the source. This script resides in the directory where you installed Storage Mirroring. By default, this file
will refresh all NICs on the source. Results will be slower if you are refreshing multiple NICs.
• Set the NICs on the source not to update the DNS server(s). This can be managed from the DNS server
or on the server’s Advanced properties DNS tab for the NIC.
• In a non-domain environment when using Replicate NT security by name, enable this feature on both
servers since replication will occur in both directions.
• Before a failover occurs, ensure that identical accounts and shares are available for each user on the
stand-in server.
• In the monitor settings, if you choose to failover and failback the Hostname in an Active Directory
environment, you will need to understand how Service Principal Names work. For more information,
see Active Directory and Service Principal Names on page 12.
In addition, you should consider the following to ensure that the clients have full access after a failover or
failback:
• On the clients, you should run the command
ipconfig /flushdns after failover to refresh the DNS
cache.
• After failover, you many need to disconnect, then re-connect clients to current mapped shares.
• After failover or failback, clients may need to press F5 to refresh the share path. If this does not work,
client users should logout and log back in to regain share information and to gain immediate access to
shares.
NOTE: There is a possibility of data loss when completing a restore and not failing back immediately
when the clients are still connected to the target. If you have failed over and are completing a restore, it is
best to failback immediately.
Active Directory and Service Principal Names
Service Principle Names (SPNs) are properties of Active Directory (AD) computer accounts. When an AD
client attempts to use a network share, it checks AD to see what computer is associated with the requested
host name. When the client attempts to establish a session with the server, a message stating “Login