5.2.1 Prerequisites for a logical host environment
When operating JP1 in a logical host environment, you must satisfy the following prerequisites for a logical IP address and
shared disk:
Table 5‒2: Prerequisites for a logical host environment
Logical host
component
|
Prerequisites
|
Shared disk
|
-
A shared disk must be used that can be taken over from the primary server to the secondary server.
-
The shared disk must be assigned before JP1 is started.
-
The assignment of the shared disk must not be canceled while JP1 is running.
-
The assignment of the shared disk must be canceled after JP1 is stopped.
-
The shared disk must be locked so that multiple servers do not inadvertently use it.
-
Files must be protected using a file system with the journal functionality or other measures so that the files will not be
lost due to a system failure.
-
Failovers must guarantee that the contents of all files are taken over correctly.
-
Failovers must be forced to occur even when a process is using the shared disk during failover.
-
Clustering software must be responsible for recovery upon the detection of any failure on the shared disk so that JP1 does
not need to perform recovery. Clustering software must issue a start or stop request to JP1 if it is necessary to start or
stop JP1 as part of recovery.
-
Use storage connected via SCSI or SAN (Fibre) as the shared disk for JP1/Base. Do not use network connection disks (NFS, NAS,
and others) as the shared disk. As performance varies depending on the disk, carry out investigations in advance.
|
Logical IP address
|
-
Communication must be performed using a logical IP address that can be taken over.
-
The logical IP address must be uniquely determined from the logical host name.
-
The logical IP address must be assigned before JP1 is started.
-
The logical IP address must not be deleted while JP1 is running.
-
The correspondence between the logical host name and logical IP address must not be modified while JP1 is running.
-
The logical IP address must be deleted after JP1 is stopped.
-
Clustering software must be responsible for recovery upon the detection of a network failure so that JP1 does not need to
perform a recovery. Clustering software must issue a start or stop request to JP1 if it is necessary to start or stop JP1
as part of recovery.
|
If any of the above requirements are not satisfied, JP1 might malfunction. For example:
-
If data written from the primary server corrupts upon a failover
JP1 might encounter a problem, such as an error, lost data, or a failure in starting up.
-
If no recovery is performed when a LAN board fails
A communication error occurs, preventing JP1 from operating normally until clustering software switches the LAN board or failover
to another server occurs.
To Page Top