Copyright © 2010, Juniper Networks, Inc.
NOTE: For additional redundancy, we recommend that you install at least
two additional heartbeat network connections. This installation protects
against the heartbeat network connection from being the single point of
failure for the entire system. For example, if a shared disk setup is used, in
case one of the heartbeat network connections goes down, both servers
would not consider the other server as dead, thus mounting the shared disk
simultaneously, resulting in a corrupted file system. If you choose to install
two network cards, we recommend that you use one dedicated interface for
heartbeat communications, in addition to one for network communications.
In the event of a process failure on the primary server, the primary server proceeds as
follows:
Shuts down all local server processes.
1.
Synchronizes all information to disk.
2.
Unmounts the shared partitions (if using a shared disk).
3.
Signals to the secondary server that it is done shutting down.
4.
The HA process in the primary server then enters an ERROR mode, and stays in that mode
until you manually restart the HA Server.
NOTE: You cannot start or stop the Device Server and GUI Server processes
manually in an HA configuration. You must use the HA Server to control these
services.
NOTE: To prevent the server from rebooting in a HA configuration that uses
shared disks, you must ensure that none of the shared files are in use before
stopping the HaSvr process. If these files are in use (for example, by a vi or
tail command), then the configured file system unmount command will fail,
causing the server to reboot.
Restoring Connections
In the event that the GUI Server fails over, the Device Server detects this status and
automatically reconnects to the secondary GUI Server.
If you are attempting to connect to the GUI Server using the User Interface, you must
enter the secondary server IP sddress to reconnect to the new GUI server IP address.
NOTE: After failover, it will take some time for the standby management
system to become fully active with the replicated database. For large
networks, this can take up to 10 minutes.
Chapter 5: Installing NSM with High Availability
75
Need help?
Do you have a question about the NETWORK AND SECURITY MANAGER 2010.4 - INSTALLATION GUIDE REV1 and is the answer not in the manual?