Problem
You created an on-host integration with New Relic, but you are not seeing data in the New Relic's Infrastructure UI.
Solution
To troubleshoot and resolve the problem:
- Check to see if the integration is activated and reporting. Go to one.newrelic.com > Infrastructure > Third-party Services and look for the integration name under Active integrations.
- Run New Relic Diagnostics to check for common issues and validate your configuration files.
- Verify that your on-host integration meets New Relic's integration requirements:
- Troubleshoot integration requirements
-
If you are not receiving data from your on-host integration, verify that your integration follows these requirements.
Integration requirements Comments Environment Make sure your environment meets the Integrations requirements listed for your specific on-host integration. Installation Make sure the installation procedures for your specific on-host integration completed successfully. Configuration file Check your configuration file:
- Use a validator such as yamllint to verify it is a valid YAML file.
- Verify that the file is in the correct location.
Definition file Check your definition file:
- Use a validator such as yamllint to verify it is a valid YAML file.
- Verify that the header fields have the required format.
- Verify that the prefix for inventory set in the definition file is no more than two levels deep.
- Verify that the file is in the correct location.
Metric sets Verify that the integration:
- Does not generate metric sets with more than 120 key-value pairs.
- Does not generate more than 1000 metric sets.
JSON payload Verify that the integration does not generate a single JSON payload of more than 5 MB.
- After ruling out common problems with integration requirements, follow the more in-depth troubleshooting procedures for error logs and integration loading:
- Check the integration log file for error messages
-
Recommendation: Configure a log file in the infrastructure agent configuration. This helps separate the types of errors so you can spot integration errors more easily.
When there is an error loading or running your integration, the infrastructure agent adds an error message to the log file. Errors are logged even if
verbose
mode is disabled.- Check the log file for lines that include
"level=error"
. - If there are no error messages, check whether the infrastructure agent is loading the integration correctly.
- Check the log file for lines that include
- Check that the integration is loading correctly
-
To verify whether the infrastructure agent is loading the integration correctly, try these steps:
- Enable
verbose
mode in the infrastructure agent configuration. - Restart the infrastructure agent.
-
Verify that the first lines of the log file contain two messages:
"loaded plugin" plugin=<YOUR_INTEGRATION_NAME> "found plugin config file" plugin-path="<YOUR_INTEGRATION_CONFIG_PATH>"
- Check for errors just before or after these lines that indicate a problem with loading.
The agent schedules data to be sent based on the schedule set in the definition file. Look in the log file for the integration's JSON.
JSON payload example
nginx-server-metrics: {\"name\":\"com.newrelic.nginx\",\"protocol_version\":\"1\",\"integration_version\":\"0.5.0\",\"metrics\": [{\"event_type\":\"NginxSample\",\"net.connectionsAcceptedPerSecond\":0.10344827586206896,\"net.connectionsActive\":1, \"net.connectionsDroppedPerSecond\":0,\"net.connectionsReading\":0,\"net.connectionsWaiting\":0,\"net.connectionsWriting\":1, \"net.requestsPerSecond\":0.10344827586206896,\"software.edition\":\"open source\",\"software.version\":\"1.4.6 (Ubuntu)\"}],\"inventory\":{},\"events\":[]}
- Enable
- Check resource consumption
- If integrations stop reporting, it may be due to insufficient system resources. We recommend to have no more than 100 instances of an integration running at the same time.