'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH
0096-0699-A
8.
Initiate a failover to Server B in the NHAS console. In other words, we will cause Server
A to fail and transfer all network services to Server B. For the purposes of this
illustration, cause Server A to fail by powering off the server. Initiating a failover by
pressing F at the console requires that the NHAS software be paused on the downed
server to prevent immediate recovery. Powering the server off ensures that the failover
will be successful and sustained.
At this point, Server B will execute the FAIL2.NCF file which will assign Server A’s
former secondary IP address as a secondary IP address handled by Server B.
Throughout the process, the client ping should be able to reach 123.45.67.11. After the
nearly instantaneous transfer is complete, Server B alone will respond to any requests
sent to both 123.45.67.11 and 123.45.67.22
In essence, the client is now talking to Server B exclusively. If a failback to Server A is
initiated at the console, Server B executes the CO2.NCF script file, which deletes the
123.45.67.11 secondary IP transferred from Server A. At the same time, Server A will
process the CO1.NCF file and, once again, handle requests for IP address 123.45.67.11.
With the failback complete, Server B will handle only requests to 123.45.67.22.
Notes on Failover
As an emerging technology, clustering presents unique challenges to legacy applications and
environments. While none of these challenges is prohibitive, certain steps or precautions are often
necessary to accommodate the failover process.
Server Applications and the SYS: Volume
Some applications, such as Netscape Fast Track Server, write the executable files to the server’s
SYS: volume. In the event of a failover, the surviving server would no longer be able to access
the files on the downed server’s SYS: volume. If your ProLiant cluster runs applications that
write to the server’s SYS: volume, you must install the executables for that application on both
servers. In addition, you must create a script that will launch any such applications on the
surviving server after a failover.
Client Drive Mappings with NDS and Novell Client 32
If Novell Client32 and NDS are being used, the clients will remain attached to the network during
a cluster failover. Network drives that were mapped attached to drives previously owned by the
failed server, though, will be broken. The existing Novell Client32 technology does not support
automatic failover drive mappings. A failed-over shared drive can be re-mapped in Windows by
using Windows Explorer as follows:
1.
Right click on the Start button.
2.
Select Explore.
3.
Select Map Network Drive from the Tools menu. Then select a drive letter and specify the
path to the data to be mounted as a drive on the client.
4.
Finally, choose whether you want to reconnect this drive mapping. Then select OK.
If the client applications are IP only, and the virtual IP addresses are transferred at failover to
the surviving server, then these applications will reconnect after the failover is complete.
