Proactive insights, news and tips from Oracle Fusion Middleware Support. Learn Oracle from Oracle.

Dynamic Monitoring Service (DMS) Configuration Dumping and CPU Utilization

There was recently a report of CPU spikes on a system that were occuring at precise 3 hour intervals. Research revealed that the spikes were the
result of the Dynamic Monitoring Service generating a metrics dump and writing it under the server 'logs' folder for every WLS server in the
domain. This blog provides some information on what this is for and how to control it.

The Dynamic Monitoring Service is a facility in FMw (JRF to be more precise) that collects runtime data on the components deployed to WebLogic. Each
component is responsible for how much or how little they use the service and SOA collects a fair amount of information. To view what is collected on any
running server you can use the following URL, http://host:port/dms/Spy and login with admin credentials.

DMS is essentially always running and collecting this information in the runtime and to protect against loss of this data it also runs
automatic backups, by default at the 3 hour interval mentioned above. Most of the management options for DMS are exposed through WLST but these settings are
not so we must open the dms_config.xml file which can be found in DOMAIN_HOME/config/fmwconfig/servers/<server_name>/dms_config.xml.

The contents are fairly short and at the bottom you will find the following entry:


    <dump intervalSeconds="10800" maxSizeMBytes="75" enabled="true"/>


The interval of 10800 seconds corresponds to the 3 hours and the maximum size is 75MB. The file is written as an archive to DOMAIN_HOME/servers/<server_name>/logs/metrics.
This archive contains the dump in XML format.

You can disable the dumps all together by simply setting the 'enabled' value to 'false' or of course you could modify the other parameters to suit your needs. Disabling the
dumps will NOT impact DMS collections or display at runtime. It will only eliminate these periodic backups.

Be the first to comment

Comments ( 0 )
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.