Application Code probes

An application code probe monitors agent schedules and the resources that agents use. For example, using an application code probe, you can identify agents that use excessive resources, that run longer than they should, and that fall behind schedule. In addition, probes can suggest solutions to resolve problems.

Application probes monitor server-based scheduled agents that the Agent Manager runs and server-based Web agents that the Domino® HTTP process runs. On a daily basis, each application probe identifies the 100 most problematic agents in each probe subtype.

The Application Code -- Long running agent probe, which identifies potential run-away agents, generates a ranked list of agents that have run the longest. The list of 100 daily problems is restarted each day at midnight for HTTP probes and at a cache refresh time for Agent Manager probes, as specified in the Server document. The minimum configuration for scheduled agents run by the Agent Manager is five minutes, which corresponds to the polling interval. The minimum configuration for We agents run by the HTTP process is one minute, which corresponds to the polling interval. The polling interval is not configurable. When the probe is enabled, all agents are monitored.

The Application Code -- Agents ranked by memory usage probe tracks memory use of LotusScript® and Java™ agents. The ranking for the same agent may differ when the agent runs in Agent Manager as compared to when it runs in HTTP. If HTTP is configured to run concurrent agents, the memory is shared by concurrently-executing agents. Therefore, the threshold for rankings depends, in part, on the current server setting for the maximum number of HTTP active threads, whether the Web agents are configured to be executed concurrently, and on the memory used by the agent. When the probe is enabled, all agents are monitored.

Specify the maximum number of threads in the Number of Active Threads field on the Internet Protocols > HTTP tab of the Domino Server document. Specify the concurrent agents setting in the Run web agents concurrently field on the Internet Protocols > Domino Web Engine tab of the Domino Server document.

Note: HTTP probes (Web agent probes) are not available for IBM® AIX®, Linux™ and Linux on IBM zSeries®.

Another type of probe does not require any configuration on your part, is always running, and identifies several error conditions. This probe does not limit its error reporting to 100 errors. It monitors the following conditions:

Agent security errors
Disabled agents for the design update task
Agents that terminate abnormally when the maximum time to run is exceeded
Full-text search operations performed on databases that are not full-text indexed

This table contains the names and descriptions of the Application Code probes that you can define.

Table 1. Application Code probes
Application code probe name	Description
Application code -- Agents behind schedule	Identifies which agents have fallen behind schedule by the greatest amount of time for the current day. The minimum configuration for this probe is five minutes, which corresponds with the polling interval. This probe applies to agents run by the Agent Manager.
Application code -- Agents ranked by CPU usage	Generates a list of the 100 agents that use the greatest amounts of CPU for the current day. This probe has a relatively high overhead. When the probe is enabled, all agents are monitored. This probe applies to agents run by the Agent Manager and Web agents.
Application code -- Agents ranked by memory usage	Tracks memory use of LotusScript and Java agents. This probe applies to agents run by the Agent Manager and Web agents.
Application code -- Long running agents	Identifies potential run-away agents, generates a ranked list of agents that have run the longest. This probe applies to agents run by the Agent Manager and Web agents. The minimum configuration for Web agents is one minute; the minimum configuration for scheduled agents run by the Agent Manager is five minutes.