Chapter 2   Monitoring Server Activity


This chapter describes the monitoring service provided by iASAT. This service allows you to chart various attributes of the Executive, Java, C++ and Bridge server processes.

The following topics are included in this chapter:

• Monitoring iPlanet Application Server

• Receiving Event Notification

Monitoring iPlanet Application Server

---

iPlanet Application Server Administration Tool (iASAT) provides a monitoring service that lets you chart the activity of the Executive, Java, C++ and Bridge servers that make up iPlanet Application Server. You can also log the information to a file. By graphically representing this server activity or recording the data in a file, you can track and review the performance of an application server or group of servers and make adjustments to improve performance. For example, if you add more memory to the application server or deploy a new application, you may want to monitor the performance of the application server to see the impact of these changes.

iPlanet Application Server's monitoring service polls the application server at designated intervals. This saves server resources because the server updates the information being monitored at the specified interval instead of updating it continuously. You can specify this interval in the Monitoring window. For information about setting the interval time, see To Change a Process Data Plot.

The monitoring window "pops out" from the Administration Tool when you click a process to monitor. This detached window enables you to monitor server activity in a separate window while continuing to perform other administrative tasks using the Administration Tool.

This section describes the following topics:

• Monitoring Process Attributes

• To Log Process Data to a File

• To Change a Process Data Plot

• To Remove a Process Data Plot

Monitoring Process Attributes

The server activity, or attributes, you can chart varies according to which server, or process, you are monitoring.

The Executive Server (KXS) process is responsible for managing and hosting the system-level services, such as the load-balancing service, and for delegating requests to one of the application processes, either the Java server, or C++ server depending on the language in which the application component is written.

You can chart the following attributes of the Executive Server process:

Table 2-1    Executive Server Monitoring Attributes  

Executive Server Process Attribute (KXS)	Description
CPU load	The amount of load on the CPU on which this Executive Server process is running, as calculated by the load balancing service.
Disk input and output	The rate of Read and Write operations issued by the system on which this Executive Server is running, as calculated by the load balancing service.
Memory thrash	The number of pages read from or written to the hard disk drive to resolve memory references to pages that were not in memory at the time of the reference.
Requests queued	Number of requests currently waiting in the queue for processing.
Cached results	Number of entries stored in the result cache.
Average execution time	Average amount of time for the Executive Server process to execute a request.
Requests/interval	Number of new requests received since the last polling.
Total requests	Total number of requests the process has received starting up (This value is reset to 0 upon server or process start-up.). For the executive process, this corresponds to the total number of requests the server has executed across all server processes.
Current requests	The number of requests currently being processed by the server; includes all requests dispatched and being processed in the KJS/KCS engines.
Requests waiting	Number of queued requests waiting to be serviced.
Requests ready	Number of queued requests ready to be serviced.
Current Requests Threads	Number of request threads allocated by the process (includes both idle threads and threads actively processing requests). Note that this number cannot exceed Maximum Threads, or fall below Minimum Threads configured for this process. These values are set in the General window.
Requests Threads Waiting	Number of requests threads available to execute new incoming requests. This number will be a subset of the Current Requests Threads monitoring attribute.
Total Threads	Number of threads being used by the process.
Bytes sent/interval	Number of new bytes sent since the last polling.
Bytes received/interval	Number of new bytes received since the last polling.
Current Sessions	Number of current sessions being handled.

---

Note	If you monitor CPU load, disk input and output, or memory thrash, you must specify the intervals at which the statistics for these process attributes are updated. To set the intervals, select Load Balancing. Choose User Defined Criteria Load Balancing, then click the Advanced Settings tab.

---

The Java Server (KJS) and C++ Server (KCS) processes are responsible for hosting application elements, depending on the language in which the element is written. The Java Server hosts application components written in Java, and the C++ Server hosts components written in C++. In addition, the Corba Executive Server (CXS) or Bridge process allows for independent Java clients (Rich Clients) to communicate directly to Enterprise JavaBeans hosted on a Java Server. For more information about Rich Client, see the Developer's Guide (Java).

You can chart the following attributes of the Java, C++ and Bridge Server processes:

Table 2-2    Java, C++ and IIOP Bridge Server Monitoring Attributes  

Java/C++ and Bridge Server Processes (KJS, KCS and CXS) Monitoring Attributes	Description
Average execution time	Average amount of time for the process to execute a request.
Requests/interval	Number of new requests received since within the interval.
Total requests	Total number of requests the process has received since the last start-up. This value is reset to zero upon server or process start-up.
Current Requests	The number of requests currently being processed by this process.
Requests Waiting	Number of queued requests waiting to be serviced.
Requests Ready	Number of queued requests ready to be serviced.
Current Requests Threads	Number of request threads allocated by the process (includes both idle threads and threads actively processing requests). Note that this number cannot exceed Maximum Threads, or fall below Minimum Threads configured for this process. These values are set in the General window.
Request Threads Waiting	Number of requests threads available to execute new incoming requests. This number will be a subset of the Current Requests Threads monitoring attribute.
Active data connections	Number of currently active data connections.
Cached data connections	Number of currently cached data connections.
Queries/interval	Number of queries executed within the interval.
Trans committed/interval	Number of transactions committed within the interval.
Trans rolledback/interval	Number of transactions rolled back within the interval.
Total Threads	Number of threads being used by the process.
Bytes sent/interval	Number of new bytes sent since the last polling.
Bytes received/interval	Number of new bytes received since the last polling.

For each process, you can chart one or more attributes. You can also simultaneously chart the attributes of several application servers, if you have a multiple-server enterprise.

To Monitor Process Attributes

To monitor process attributes, perform the following tasks:

1. On the iASAT toolbar, click Monitor to open the Monitor window.

2. In the left pane of the Monitor window, click the process whose attributes you want to chart, as shown in the following figure:

   
   
   

   A separate monitoring panel pops out of iASAT, as shown in the following figure:

   
   
   

3. In the right pane of the monitoring window in iASAT, click Add Plot located at the bottom of the window.

   The Add Plot dialog opens for you to specify the attributes to monitor for the highlighted process.

   
   
   

4. In the Attribute drop-down list, select the attribute to chart.

5. From the Scale drop-down list, choose the ratio (scale) at which to plot the attribute.

   Values range from 10:1 to 1:1,000,000. A scale of 10 to 1 (10:1) indicates that 10 units will be plotted on the Process Monitor window for each attribute count.

6. From the Color drop-down list, choose a color to represent the process attribute on the chart.

7. Repeat steps 2 through 6 for each process or attribute you want to chart.

   Each process attribute that you choose to chart is displayed in the Monitor window.

   
   
   

8. In the Time Interval drop-down list, select the interval at which you want to update the Monitor Plot window.

   This setting applies to all process attributes displayed in the Monitor window.

To Log Process Data to a File

Once you begin monitoring a process attribute, you can send data collected by the monitoring service to a file.

To log process data to a file, perform the following steps:

1. On the iASAT toolbar, click Monitor to open the Monitor window.

2. In the left pane of the Monitor window, click the process whose data you want to chart.

3. In the right pane of the Monitor window, click Options.

   The following dialog box appears:

   
   
   

4. Click the Log to File checkbox to enable the logging service.

5. In the File Name text field, enter the name of the file where data is to be written.

6. Click OK to save your changes.

To Change a Process Data Plot

Once an attribute data plot is specified for a process (KCS, KJS, and KXS), you can adjust the plot using the Attribute, Color, and Scale drop-down boxes.

To change the way a process attribute is plotted, perform the following steps:

1. On the iASAT toolbar, click Monitor to open the Monitor window.

2. In the right pane of the window, select a process row, whose attributes you want to change.

3. To change an attribute, click the Attribute column and choose a new attribute to plot from the drop-down list.

4. Similarly, click the Color and Scale columns to change how the attribute will be plotted.

To Remove a Process Data Plot

If you no longer want to plot an attribute for a process, you can remove it from the plot.

To remove a process plot, perform the following steps:

1. Click Monitor on the iASAT toolbar to open the Monitor window.

2. In the left pane of the Monitor window, choose a process, to display the process rows in the right pane.

3. In the right pane of the Monitor window, select a process whose attribute you want to remove.

4. Click Remove Plot.

   The attribute is removed from the Monitor window.

Receiving Event Notification

---

Event notification is useful when you cannot actively monitor an iPlanet Application Server. This passive monitoring system is activated only in critical circumstances, such as when a process has failed.

You can set the system to alert one or more concerned parties through email when a critical situation arises by supplying the email addresses of those you want to alert. In addition, you can specify a script that will run automatically when certain events occur.

This section includes the following topics:

• About Events

• To Configure Email Notification for an Event

• To Specify an Event-Invoked Script

About Events

You can specify an individual to notify or a script to run for the following critical events:

• Executive Sever (KXS) goes down

• Java Server (KJS) goes down

• C++ Server (KCS) goes down

• Process auto restarts exceeded

• Abnormal Cluster is detected

This section includes details for the following topics:

• What Do I Do When a Server Goes Down?

• What Do I Do When Restarts Are Exceeded?

• What Do I Do When an Abnormal Cluster is Detected?

What Do I Do When a Server Goes Down?

If one or more of the Executive Server, Java Server, or C++ Server processes go down, the Administrative Server attempts to restart each process. If the process cannot be restarted by the Administrative Server, the application stops running and can result in lost transactions.

Recurring failures are usually attributed to problems within the application code, but other failures can also happen. Regardless of what causes a process to fail, it is useful to be notified immediately.

If the process restarts, investigate the cause of the failure to determine whether adjustments can be made to prevent future failures. If the process does not restart, look at the log to find the cause of the failure.

What Do I Do When Restarts Are Exceeded?

You can also be notified when the Administrative Server has exceeded the number of times it has been set to restart a process. The maximum engine restarts value is set on the Server tab of the General window.

Increase the Administrative Server restart option, if it is low, and determine the cause of the process failure.

What Do I Do When an Abnormal Cluster is Detected?

You can also be notified when an abnormal cluster condition has been detected. Within a normal operating cluster there is one sync primary iPlanet Application Server that is the primary data store, with which all other cluster members communicate for the latest distributed data information. An abnormal cluster is where a dual-primary or a no-primary condition has been detected.

Mark the Restart in case of abnormal cluster checkbox on the Cluster tab of the General window. iPlanet Application Server will re-start an appropriate process so that one (and only one) sync primary is present in the cluster. For more information about clusters, see Managing Distributed Data Synchronization.

To Configure Email Notification for an Event

To send an email notification for an event, perform the following steps:

1. On the iASAT toolbar, click Events to open the Events window.

2. From the left pane of the Events window, select the server for which you want to configure events.

3. From the right pane of the Events window, select the event or events for which you want to be notified by clicking the corresponding checkbox as shown in the following figure:

   
   
   

4. In the Email Addresses field, specify the email address or addresses of the persons you want to send notification. To use multiple email addresses, separate each email address with a semi-colon(;), as shown in the following example:

   betsy@doghouse.com;arland@meow.com

5. In the Mail Server field, specify the mail server through which the notification is sent. Use the following format:

   mail.company.com

6. To see the most recent events that might have been sent out for this server, click Poll for Events.

   The Poll for Event dialog box appears displaying a list of the recent events for the selected server.

   
   

   ---

   Note	When you click Poll for Events, events are consumed (that is, the events that you see are no longer included in the next set of events that are displayed).

   ---

   
   

7. Click Apply Changes to save your changes.

To Specify an Event-Invoked Script

You can configure the event notification service to run a script. The script might page the system administrator, bringing the problem to the administrator's attention, or perform any other automated task that will help keep the system running smoothly when faced with a critical event.

When a script runs, it passes an argument to indicate what type of event has occurred. For instance, the following command indicates that a Java Server (KJS) process has crashed:

/script location/ crash kjs

To configure the event notification service to run a script in response to an event, perform the following steps:

1. On the iASAT toolbar, click Events to open the Events window.

2. From the left pane of the Events window, select the server for which you want to configure events.

3. In the right pane of the Events window, mark the checkboxes against the events for which you want to invoke a script.

4. In the Script field, specify the path of the script to run. For example:

   /mydir/scripts/myscript.pl

5. Click Apply Changes to save your changes.