7
Tuning the Oracle Management Server

There may be special circumstances which require specific tuning of the Management Server to improve performance such as when the enterprise is managing a large number of nodes. Tuning can be performed by setting the Management Server configuration parameters. Refer to the sections below.

After editing the omsconfig.properties file, you must stop and restart the Management Server in order for the changes to take effect.

Setting the Ping Interval

The Management Server is designed to ping Agents on all targets' hosts on a pre-defined interval to monitor the availability of the Agent and its target host.

To manage the interval between pings, you can use the following property in the omsconfig.properties file:

oms.vdp.ping_interval=<integer; time in minutes; default 2> 

The Management Sever actually pings the Agents at intervals of 1/2 of the oms.vdp.ping_interval value. Maintaining the default value for this parameter (i.e. 2 minutes) indicates that the Management Server will ping each Agent at 1 minute intervals. For instance, if you would like to have the Management Server ping Agents every 10 minutes, set the oms.vdp.ping_interval to 20.

If the Node Up/Down event test is registered against a node, then this pinging of the Agent by the Management Server is actually used to determine the status of the node up/down event. When the Management Server pings the agent, and if the Agent or its target host is unavailable, then the node up/down event triggers with the appropriate message.

For Management Servers managing a large number of nodes (more than 64 nodes), you can adjust this parameter to provide the Management Server enough time to ping all the nodes.

Setting the Maximum Connections Out

The Management Server is designed to maintain a certain number of outgoing connections toward different Intelligent Agents simultaneously.

To manage the number of simultaneous outgoing connections, you can use the following property in the omsconfig.properties file:

oms.vdg.max_out_conns =<integer; default 64>

The default is 64. Oracle recommends that this value to be greater or equal to the number of nodes managed by the Management Server in order to avoid performance degradation. If the maximum number of managed nodes is less than 64, Oracle recommends keeping the default value. If the maximum number of managed nodes is more than 64, set the value to the number of nodes.

Setting the Maximum Connections In

The Management Server is designed to maintain a certain number of incoming connections simultaneously.

To manage the number of simultaneous incoming connections, you can use the following property in the omsconfig.properties file:

oms.vdg.max_in_conns=<integer; default 32>

To prevent performance degradation, it is recommended that you set this value to be half the value of the oms.vdg.max_out_conns.

Setting the Management Server Retry Interval

If the Management Server and the repository server lose connection, the Management Server can be configured to try to re-establish the connection to the repository server. The Management Server is designed to retry at a pre-defined intervals with these two parameters:

• oms.repository.connect_timeout

• oms.repository.connect_numTries

To specify the interval between retries, use the following properties in the omsconfig.properties file to set the retry interval:

oms.repository.connect_timeout=<time in seconds>
oms.repository.connect_numTries=<integer>

oms.repository.connect_timeout is the maximum time (in seconds) the Management Server will wait for the repository database to start up.

oms.repository.connect_numTries is the maximum number of tries the Management Server will make to connect to the repository at the startup of the Management Server service.

The management server retry interval for connection to the repository is calculated as the value of oms.repository.connect_timeout divided by the value of oms.repository.connect_numTries.

By default, the properties are set as follows:

oms.repository.connect_timeout=120
oms.repository.connect_numTries=12

The Management Server will try to establish a repository connection at startup at 10 second intervals for a total of 12 times.

Smoothing Over Temporary Network Failures

The Management Server tries to ping all nodes every interval in order to detect whether a node is up or down.

If it detects that it was able to successfully ping a node in the previous interval and not in the present interval, it would mark that node as down.

The ping can be made more tolerant to temporary network outages by the following parameters.

oms.vdg.conn_retries = (integer, default 1). 

This parameter specifies the number of retries the Management Server will try to establish a connection either for pinging or for other communication. The Management Server defaults to 1. In other words, it will not retry any connection.

For networks prone to temporary outages, set it to 2 or 3.

oms.vdg.conn_retries_delay = (integer, default 5)

This parameter specifies the amount of time in seconds the Management Server will wait between retries to establish a connection.

This parameter needs to be set to a value depending on the duration of network outages.