Job Management Partner 1/Automatic Job Management System 3 Administration Guide
When a cluster configuration is used, JP1/AJS3 operates in a logical host environment to support a failover. For JP1/AJS3 to operate in a logical host environment, the cluster software must manage allocation, deletion, and monitoring of the shared disk and logical IP address correctly.
- Cautionary note
- Depending on the system configuration and environment settings, even the cluster software supported by JP1/AJS3 may be unable to satisfy the prerequisites described in this subsection. If it does not satisfy the prerequisites, change the system configuration and environment settings so that the prerequisites are satisfied.
- Organization of this subsection
- (1) Prerequisites for a logical host environment
- (2) Prerequisites for a physical host environment
- (3) Scope supported by JP1/AJS3
- (4) Requirements for a logical host name
(1) Prerequisites for a logical host environment
The following are the prerequisites for using JP1/AJS3 in a cluster system.
- The cluster software must be able to control prerequisites 2 to 4 below.
- A shared disk that enables data to pass from the primary node to the secondary node must be available.
The details are as follows:
- The shared disk must be allocated before JP1/Base and JP1/AJS3 start.
- The shared disk must not be deallocated while JP1/Base and JP1/AJS3 are running.
- The shared disk must be deallocated after JP1/AJS3 and JP1/Base stop.
- The shared disk must be locked so that it is accessible only from the primary node.
- The files must be protected by the journal functionality of a file system so that they will not be lost if the system goes down or another problem occurs.
- In a failover, the integrity of the information written to the files must be assured and that information must be inherited.
- Forced failover must be possible if a process is using the shared disk.
- If an error is detected on the shared disk, the cluster software must control recovery measures, which must be transparent to JP1/Base and JP1/AJS3. If JP1/Base and JP1/AJS3 need to be started or stopped in extended recovery measures, the cluster software must issue the start and stop requests to JP1/Base and JP1/AJS3.
- For logical IP addresses, the following conditions must be satisfied:
- Communication using a logical IP address that can be inherited must be possible.
- A unique logical IP address must be obtained from a logical host name.
- A logical IP address must have been assigned before JP1/Base and JP1/AJS3 start.
- A logical IP address must not be deleted while JP1/Base and JP1/AJS3 are running.
- The correspondence between logical host names and logical IP addresses must not change while JP1/Base and JP1/AJS3 are running.
- The logical IP address must be deleted after JP1/AJS3 and JP1/Base have stopped.
- When a network error is detected, the cluster software must control recovery measures, which must be transparent to JP1/Base and JP1/AJS3. If JP1/Base and JP1/AJS3 need to be started or stopped in extended recovery measures, the cluster software must issue the start and stop requests to JP1/Base and JP1/AJS3.
- For concurrent execution, a separate IP address must be assigned to each logical host.
- A logical host name must be set in the hosts file and on the name server so that TCP/IP communication is possible. When the DNS is used, host names that are not in FQDN format must be available.
- JP1/Base, which is a prerequisite program for JP1/AJS3, must be in an environment in which a cluster system can be used.
If the above prerequisites are not satisfied, a problem may occur during operation of JP1/AJS3. For example, the following problems may occur.
- If data written on the primary node is damaged when a failover occurs:
A problem such as an error, data loss, or start failure occurs and JP1/AJS3 cannot operate normally.
- If recovery is not performed even when an error occurs in the NIC:
A communication error occurs and JP1/AJS3 does not operate normally until the cluster software switches over the NIC or performs a failover to the other server occurs.
(2) Prerequisites for a physical host environment
The following table shows the prerequisites for using JP1/AJS3 in a physical host environment. These prerequisites must be satisfied even when you want to execute only JP1/AJS3 in a logical host environment.
Table 11-1 Prerequisites for a physical host environment
Physical host component Prerequisites Server main unit
- Two or more servers are used to create a cluster system.
- The CPU is capable of handling the expected processing even when multiple logical hosts are concurrently activated.
- The servers have sufficient memory for handling the expected processing even when multiple logical hosts are concurrently activated.
Disk
- The files in the disk are protected by the journal functionality of the file system so that they will not be lost if the system fails.
Network
- The IP address for the host name (acquired by the hostname command) can be used for communication (the cluster software does not disable communication).
- The mappings between host names and IP addresses are not changed while JP1/AJS3 is operating (the cluster software or name server does not change the mappings).
- In Windows, the NIC for the host name has the highest priority in the settings for network binding (the highest priority is not given to other NICs such as a NIC for the heartbeat LAN).
OS and cluster software
- The versions of the OS cluster software are supported by JP1/AJS3.
- The patches and service packs prerequisites for JP1/AJS3 and cluster software have already been applied.
- Each server has the same environment so that the same processing can be executed if a failover occurs.
(3) Scope supported by JP1/AJS3
When JP1/AJS3 operates in a logical host environment, the scope supported by JP1/AJS3 is its own operations only. JP1/AJS3 does not control a logical host environment.
If the prerequisites for the logical and physical host environments described above are not satisfied, or if there is a problem with the control of the logical host environment, any problem with operation of JP1 is also out of the supported scope. In this case, the problem needs to be handled on the cluster software or OS that controls the logical host environment.
- Cautionary notes
- In an environment where the local host cannot obtain the IP address from the local host name, you cannot execute jobs (standard jobs, action jobs, event jobs, or custom jobs), commands of the job execution control, and commands in a queueless job execution environment.
- If a logical host name is the same as the physical host name, you cannot use the queueless job execution facility or the definition pre-check function.
(4) Requirements for a logical host name
The following are the requirements for logical host names.
- A logical host name must be set in the hosts file and on the name server so that TCP/IP communication is possible. When the DNS is used, host names that are not in FQDN format must be available.
- The logical host names must be handled in JP1/Base, which is a prerequisite program for JP1/AJS3.
For details, see the Job Management Partner 1/Base User's Guide.
By specifying logical host names, you can start JP1/AJS3 services and execute commands on each logical host. Use either of the following methods to specify a logical host name:
- JP1_HOSTNAME environment variable
- Logical host specification option of a command (normally, the -h option)
For details, see the explanation of commands in 2. Commands in the manual Job Management Partner 1/Automatic Job Management System 3 Command Reference 1 or 2. Commands Used during Setup in the manual Job Management Partner 1/Automatic Job Management System 3 Command Reference 2.
If the logical host specification option is not specified, the command is executed with the logical host name set in the JP1_HOSTNAME environment variable. Note that the variable setting JP1_HOSTNAME="" will be ignored.
Copyright (C) 2009, 2010, Hitachi, Ltd.
Copyright (C) 2009, 2010, Hitachi Solutions, Ltd.