Job Management Partner 1/Software Distribution Client Description and User's Guide

2.4.2 Package relay system

This section describes the role of the relay system in the package distribution process.

Organization of this subsection

(1) Checking for a distribution job

(2) Polling the managing server

(3) Displaying the distribution status of packages

(4) Retaining and deleting packages

(5) Storing and deleting jobs

(6) Managing the startup status of clients

(7) Notifying the distribution results

(1) Checking for a distribution job

The relay system executes the rdsft -a command to check the managing server whether there is a distribution job for the local UNIX machine.

If there is a distribution job, the package is distributed according to the contents of the job. If there is no distribution job, the process ends. When the distribution process is executed, the program name of the distributed package will be output to the log file.

If an error occurs during processing of multiple distribution jobs for the local UNIX machine, one of the following will occur:

If an error occurs during installation
That distribution process is canceled, and the processing of another distribution process is executed.

If an error occurs during the communication with the managing server
All the processes are canceled immediately.

JP1/Software Distribution Client (relay system) can also periodically check whether there is a distribution job from JP1/Software Distribution Manager (polling). By setting to execute polling in the relay system, package distribution can be performed automatically even if the managing server fails in the startup request job. For details on polling, see (2) Polling the managing server.

(2) Polling the managing server

The relay system connected to the managing server or the relay system connected to such a relay system can execute polling. To execute polling from a relay system, specify the interval of polling in WatchTimeofOrders in the settings file of the client services.

To start polling which checks whether there is a distribution job, use the rdsma command of the client services. When the rdsma command of the client services is written in the system startup shell (/etc/localrc), polling starts when the relay system starts.

Polling ends when the system stops. You can also stop polling by executing the rdsma -q command.

When polling is used, the relay system automatically communicates with an upper-level system at the interval specified in the settings file. Therefore, depending on the network conditions, the load may be excessive or extra line fee may be required. When you check distribution jobs, consider the network environment.

(3) Displaying the distribution status of packages

JP1/Software Distribution Client (relay system) can check the status of package distribution being executed from JP1/Software Distribution Manager via a local relay system. Figure 2-3 shows how distribution status is displayed at a relay system.

Figure 2-3 Display of distribution status at a relay system

(a) Setting up the environment for displaying the distribution status

The relay system that will display the distribution status requires the file that manages the distribution status (distribution situation file) in its system. To create the distribution situation file, set OutPutFileOfDeliveryStatus and OrderOfDeliveryStatus in the settings file of the client services.

(b) How to display the distribution status

To display the distribution status, use a command of the client services. Only a superuser can execute the command. The procedures are as follows:

Execute the rdsdlvse command
The feature which manages the distribution status (distribution status server facility) starts.

Execute the rdsdlv command
The distribution status is displayed.

Execute the rdsdlvse -p command
The distribution status server facility stops.

If you execute the rdsdlv -d command while the distribution status server facility is running, you can delete information of a specific distribution status. To initialize the distribution situation file, stop the distribution status server facility, and then execute the rdsdlvclr command.

The distribution status of a package for which the Install package job has been completed is managed in only one generation. Therefore, if a package is re-delivered to a destination that already contains a package, the previous generation of the distribution status is overwritten with the job distribution status that is executed last. The distribution status of completed execution is managed by distribution-destination-name + packager-type-code + package-ID. Packages with different cabinet ID, version, or generation number are handled as the same package.

(4) Retaining and deleting packages

Packages for which the package storage period at the relay system is set during packaging, and packages for which the storage period (LifeofResource) is set in the settings file are saved in storage until the limit date.

The expired packages are deleted automatically by the JP1/Software Distribution Client (relay system) when the power is first turned on. If performing deletion once a day causes any inconvenience, you can also delete the package using the rdsclr command. Also, if you use the rdsft -init command to initialize the storage, all packages are deleted all at once.

(5) Storing and deleting jobs

The relay system saves the distribution job from the managing server in the temporary job storehouse. Jobs stored in the temporary job storehouse can be checked or deleted with the rdsgroupid command. We recommend you periodically check the jobs according to the operation.

When a package is deleted from the temporary package storehouse, the job for the package in the temporary job storehouse becomes invalid. You can set whether or not to automatically delete files that have become invalid in IDCommandDeleteTiming and CommandDeletewithResource, in the settings file of the client services.

(6) Managing the startup status of clients

The relay system can control the startup status of lower-level systems (clients) as follows:

Suppressing a startup request to the client when a distribution job for a client from the managing server is received.

Starting only specific clients

Reporting client start failure to the managing server

(a) Suppressing client startup

To prevent startup request to the client when the relay system receives a distribution job for the client, create a flag file called client-start suppressing file.

The files names of the client-start suppressing files are as follows.

For systems other than HP-UX:/NETMRDS/rdsprm/.cltwup_off

For HP-UX:/var/opt/NETMDMW/rdsprm/.cltwup_off

Create this flag file as a normal file. There are no restrictions in the file mode and files size.

When this flag file is created, a job will not be executed until a connection request is made from the client. Therefore, it is necessary to periodically confirm a distribution command from the client to the relay system.

The examples below (for systems other than HP-UX) show how to create a client-start suppressing file. Use one of these methods to create the file from the prompt screen.

> /NETMRDS/rdsprm/.cltwup_off

echo > /NETMRDS/rdsprm/.cltwup_off

(b) Starting specific clients

To start specific clients, set the start suppression described in 2.4.2(6)(a) Suppressing client startup. After this, enter the client host name to start in the following file.

File name

For systems other than HP-UX:/NETMRDS/rdsprm/wuphosts

For HP-UX:/var/opt/NETMDMW/rdsprm/wuphosts

Set the file access privilege as follows:

file mode: 644 or more

owner: 0 (root)

group: 3 (sys)
File entry format

As shown below, enter a client host name per line. When multiple host names are entered on one line, only the host name in the beginning will be valid.
client-host-name 1
client-host-name 2
    :
Notes for specification

Only the host names for the systems under the local relay system can be entered in the file. Do not specify the local relay system.

Since all the contents entered into the file will be considered to be host names, do not enter anything other than host names in the file. Do not enter blank spaces or tabs before the host name.

The IP address cannot be specified.

Specify the same host name as the one that is specified during distribution, and the host name set in /etc/hosts and DNS. The entry is case-sensitive.

If the client startup suppression is not set, this file will be ignored.

(c) Reporting client-start error

When a distribution job is received from the managing server, the relay system starts the client. In this case, if the client fails to start, that error is reported to the managing server.

Job status reported to the managing server

When the relay system determines that the client failed to start, it reports the following to the managing server depending on the cause of the failure.

EXEC-ERROR: Start error

This report is displayed when the status display command is executed on the managing server. The managing server monitors the startup failure of the clients.

Creating the file for reporting client-start error

To report the startup failure of the client to the managing server, create the following file for reporting start error on the client.

For systems other than HP-UX:/NETMRDS/rdsprm/.cltstsud

For HP-UX:/var/opt/NETMDMW/rdsprm/.cltstsud

The following is an example of creating a file for reporting client-start error (for other than HP-UX). Execute the following procedure from the prompt screen.

> /NETMRDS/rdsprm/.cltstsud

echo > /NETMRDS/rdsprm/.cltstsud

Create this file as a normal file. There is no restriction on the file mode and file size.

(7) Notifying the distribution results

When the relay system is started, it automatically executes the rdssnd command, and sends the distribution results of the local UNIX machine and the distribution results reported from the client workstation to the managing server. The monitoring time of the distribution results must be set for WatchTimeofStatus in the settings file.

To stop sending the distribution results, execute the rdssnd -q command.

[Contents][Back][Next]

[Trademarks]