Before a pipeline can receive and process data, it must
be started. It is common to run multiple pipelines to increase data
throughput or process different types of source data. Use these steps
to start a pipeline or re-start a pipeline that is down.
Before you begin
- The pipeline node hosting this pipeline must have the pipeline
executable installed.
- There must be at least one pipeline configuration file configured
for use with the pipeline that you want to start. You can specify
the pipeline configuration file to use as part of the start pipeline
command. If you do not specify the name of configuration file as part
of the pipeline command, the pipeline configuration file must be located
on the pipeline node, and it must match the name of the executable
(pipeline name specified). For example, pipeline.ini.
- If you use a script to start pipelines,
make sure the script is located in the same directory where you start
the pipeline.
- If you want to route the results of processing
from this pipeline or monitor the statistics and status of this pipeline,
register the pipeline in the Configuration Console on the Pipelines tab.
You must use one of the already registered pipeline names to start
this pipeline for monitoring or routing to complete successfully.
The pipeline configuration file must contain the connection to the
application monitor for obtaining the routing files. (After successfully
receiving routing information one time, the routing information is
saved to a local routing file. If the application monitor is not available,
the pipeline can use the routing information from the local routing
file.)
- If you are using the application monitor
to monitor pipeline status and statistics, make sure the pipeline
node has an SNMP agent installed and running before you start this
pipeline.
- If the DEFAULT_CONCURRENCY system
parameter value is set to greater than 1 or if you configured the concurrency parameter
in the pipeline configuration file for the pipeline node, you can
start multiple parallel pipeline processing threads using a single
start pipeline command.
About this task
There are three steps to starting a pipeline:
- Verify that there are no other pipelines currently running
on the pipeline node which have the same name as the pipeline you
want to start. Each pipeline must have a name unique to its pipeline
node, so make sure there are no other pipelines running with the same
name as the pipeline you want to start. (The default pipeline name
is pipeline.) There are two ways to verify this:
- If you are using the application monitor to check the
status of pipelines or route the results to other systems, look at
the Pipeline Status tab in the Configuration
Console to see if there is another pipeline running that has the same
name you want to use.
- Or at a command prompt, type the following command:
pipeline -n pipelinename -lwhere pipelinename is
the name you want to use to start the new pipeline. Make sure that
this name matches the name registered in the Configuration Console
for this pipeline.
- At a command prompt, start one or more pipelines by specifying
the type the appropriate pipeline command options and parameters using
this format:
pipeline -option parameter
Note: If you are using the application monitor
for this pipeline and it has been registered in the Configuration
Console for monitoring or routing, be sure to use the -n option
as part of the pipeline start command and specify the registered pipeline
name. If the pipeline name specified does not exactly match the registered
pipeline name (including case), the pipeline status will not display
correctly on the Configuration Console Pipeline Status tab
and any routing configured for this pipeline will not be successful.
- Verify that the command worked, and the pipeline is started
and active.
- If you are using the application monitor and this pipeline
has been registered in the Configuration Console, check the Pipeline
Status tab. If the pipeline is active,
the status displays as Active.
- If your system is running on a Microsoft® Windows® platform and you are using the services
pipeline option, you can see the status of the pipeline in the Microsoft
Windows Services control panel.
- If your system is running on a UNIX® platform
and you are using the daemons pipeline option, you can type the following
command to check for running processes:
ps
-fu userid
where userid is
the identification of the user starting the pipeline.
- Or at a command prompt, type the following command:
pipeline -npipelinename -l where pipelinename is the name of the
pipeline you just started.
If the pipeline
is active, the command prompt returns Running.