By configuring application lazy start, you can release
resources that are being consumed by inactive dynamic clusters so
that other cluster instances in the cell can use these resources.
Before you begin
- Application lazy start requires that requests are routed through
the on demand router. To create an on demand router, see Creating ODRs for more information.
- Internet Inter-ORB Protocol (IIOP) and Java™ Message Service (JMS) requests cannot be
used because they are not routed through the ODR. Do
not use application lazy start on dynamic clusters that run Session
Initiation Protocol (SIP) applications.
- Create a dynamic cluster. See Creating dynamic clusters for more information. You can also configure application
lazy start when you are creating a dynamic cluster.
About this task
An
application lazy start is the activation of the
first application server instance of a deactivated dynamic cluster
when an application request arrives. You decide which applications
to deactivate and subsequently lazily start. Use application lazy
start if you have an environment in which the ratio of the number
of dynamic clusters to the number of nodes is high, and if many dynamic
clusters are not accessed for a long period. By using application
lazy start, you can increase the performance and efficiency of your
environment.
Procedure
- Edit the dynamic cluster properties. In the
administrative console, click Servers > Dynamic clusters > dynamic_cluster_name .
- Select the If
other dynamic clusters need resources, stop all instances of this
cluster during periods of inactivity option.
- Set the Time to wait before stopping instances in
minutes if the application placement controller determines that the
resource is required by some other dynamic application cluster.
Results
The application placement controller tracks the amount of
time that a dynamic cluster has been inactive. If another dynamic
cluster needs resources, the application placement controller can
stop the inactive dynamic cluster after the specified time period.
The resources that were consumed by the inactive instance are released
and made available for use by the dynamic cluster that requires additional
resources. If a request arrives for the stopped dynamic cluster instance,
the lazy start controller is activated and at least one server instance
is started. In the meantime, HTTP error code 503 (server unavailable)
is generated. The error page informs you that the requested application
is starting and the request is resubmitted shortly.
What to do next
- Application lazy start only stops dynamic cluster instances if
other dynamic clusters need resources. If you want to stop dynamic
cluster instances after the specified time period even if the resources
are not needed, configure the proactiveIdleStop custom property.
- You can configure a custom error page for the ODR that can be
used for the 503 error. The error page can include an HTTP meta refresh
tag that causes the browser to automatically send the request again
after a certain waiting period. See Configuring ODRs for more information about configuring custom
error pages.