90 Integrating IBM Tivoli Workload Scheduler Suite and Content Manager OnDemand
will require that the PULL-CPUs have sufficient disk space to store pending data
loads until they get loaded onto the OnDemand Server.
Possible candidates for the PULL-CPU are UNIX Master Domain Managers,
UNIX Backup Masters, UNIX Domain Managers, and UNIX FTA’s. These CPUs
would then manage their agents. If the PULL-CPUs are UNIX Domain Managers,
they could manage the CPUs within their domain. If you want to dedicate one
CPU to be responsible for pulling data from all its MANAGED-CPUs, it must be a
PULL-CPU.
The only task that must be performed on the MANAGED-CPUs is to launch a job
that will contain the TWSmerge log, Netman log, Audit Plan, and Database logs
(if they exist), generating a job stdlist. This information will be pulled by the
PULL-CPU later.
4.3.2 Push method
The Push method discussed in this section will use the real-time frequency for
loading the job stdlists and use batch mode for loading the TWSmerge log,
Netman log, and, when applicable, the Audit Plan and Database logs.
The Push method is one of the most efficient methods of loading logs to the
OnDemand Server because when a job completes, its job stdlist is loaded to the
OnDemand Server automatically. If this method is implemented for each UNIX
Master Domain Manager, UNIX Backup Master, UNIX Domain Manager, UNIX
Fault Tolerant Agent, and UNIX Standard Agent, each system will be responsible
for loading its logs to the OnDemand Server. As soon as the load is complete,
the job stdlists will be available for viewing or querying on the OnDemand Server.
Important: If the OnDemand Server is down and the IBM Tivoli Workload
Scheduler Agents are linked, the number and size of these files could have an
impact on the PULL-CPU.
Important: If you have a Windows Master, you will need to have it managed
by a PULL-CPU.
Chapter 4. IBM Tivoli Workload Scheduler Distributed implementation 91
Push method and real-time operations
As discussed earlier, there are two options for moving data from IBM Tivoli
Workload Scheduler to the OnDemand Server: real-time or batch mode. In the
real-time option, the job stdlists will be loaded to the OnDemand Server as soon
as the job completes, provided that the network and OnDemand Server are
available.
Real-time loading requires that the ARSLOAD program be configured as
documented in step 1.d of “Push method and additional requirements” on
page 92. The OnDemand ARSLOAD program is invoked at system startup and
must be running for the loads on the OnDemand Server to occur as soon as the
jobs complete.
Under normal operations, a PUSH-CPU will launch a IBM Tivoli Workload
Scheduler job. This job will execute the modified jobmanrc script that will initiate
scripts that start the load process when the job completes. The scripts will
determine if the OnDemand Server is available; if the server is available, the
job’s stdlist will be loaded onto the OnDemand Server.
If the network is down or the OnDemand Server is unavailable, a link to the
original job stdlist called os_type.stdlist.yyyymmdd.Opid_num.hhmm.log will be
created in the TWShome/stdlist/ondemand directory. These files will be loaded to
the OnDemand Server by the ARSLOAD program that is running in the
background when the OnDemand Server is available.
The Push method is the most efficient method for loading data, since the pushing
of the jobs stdlists is spread over the entire day.
Push method and batch mode operations
The other option for loading data to the OnDemand Server is the batch mode.
When the jobs complete, a link to the original job stdlist called
os_type.stdlist.yyyymmdd.Opid_num.hhmm.log will be created in the
TWShome/stdlist/ondemand directory.
Note: The solution provided for the Push method was designed for UNIX.
Because of the many possible options for shell or programming languages
that can be used to create and run scripts on Windows systems, this solution
did not include converting the UNIX scripts to a format that would work on
Windows. The UNIX scripts may be converted for Windows, in which case the
Windows systems (Master Domain Manager, Backup Master, Domain
Manager, and Fault Tolerant Agent) will then be able to function as
PUSH-CPU. If the scripts are converted, please see the notes in the scripts
that apply to PUSH-CPUs and replace the logic where for variable OS=win-p.

Get Integrating IBM Tivoli Workload Scheduler and Content Manager OnDemand to Provide Centralized Job Log Processing now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.