This is an interim release which addresses performance issues by temporarily disabling some features.
At times, the agent may not release database connections quickly when the code being instrumented does not properly dispose of connections. Connections may not be returned to the pool until the one minute harvest cycle is reached.
The Status Monitor diagnostics tool has the following known issues:
- Does not function correctly for IIS6 installations and applications running under sites other than DefaultWebSite.
- Shows incorrect information for some sites.
- Shows no reporting applications and an incorrect error message about IIS being in a stopped state in some cases where apps hosted on the machine are in fact reporting to New Relic.
The agent writes too many messages to the log at INFO level. This can result in log files growing too large and a possible CPU impact. Work-around: set logging level to warning or above.
In some cases, the agent may display incorrect counts for external calls.
The Original URL displayed in the UI is not correct if the application uses an HttpModule to rewrite the url.
In some cases, the agent may inject browser monitoring code in the wrong section of the page. This can cause pages to render incorrectly. Work-around: disable browser monitoring via the API for pages experiencing this issue.
The agent reads its config file more frequently than necessary causing a potential performance problem for apps with very high frequency, very fast transactions.
Auto RUM may not inject appropriately on pages where the content-type has optional parameters. This can result in no browser monitoring functionality for these pages.
Generic exceptions cannot be ignored in the UI.
Future upgrades from 2.8 to newer releases may not remove the newrelic.xml file. This may require manual cleanup.