System-Level Monitoring for AIX, Linux, OSX/MacOS & Solaris/SunOS
Here. If you are just deploying and not compiling it, please download the release instead of cloning the repo.
- Disclaimer
- Contributing
- Requirements
- Installation & Usage Overview
plugin.json
configuration- Other configurations
New Relic has Open Sourced this integration to enable monitoring of this technology. This integration is provided AS-IS WITHOUT WARRANTY OR SUPPORT, although you can report issues and contribute to this integration via GitHub. Support for this integration is available with an Expert Services subscription.
We'd love to get your contributions to improve the Unix Monitor! Keep in mind when you submit your pull request, you'll need to sign the CLA via the click-through using CLA-Assistant. If you'd like to execute our corporate CLA, or if you have any questions, please drop us an email at [email protected].
- A New Relic account.
- As this monitor posts 'custom events' to Insights, your account must either have an active Insights trial or be a paid Insights account.
- A Unix server that you want to monitor
- Java JRE/JDK v1.6 or later
- Network access to New Relic (proxies are supported, see details below)
- For Dashboard Installation:
curl
orwget
installed
- AIX 7.x
- HP-UX 11.x
- Linux - All sorts, including on ARM processors (such as Raspberry Pi) and z/Linux
- OSX / MacOS 10.9 ('Mavericks') and above
- Solaris/SunOS 10.x and 11.x
- Download the latest version of the Unix Monitor here.
- Copy, gunzip & untar the latest release onto the Unix server that you want to monitor
- Set account ID, keys and other settings in
config/plugin.json
- OPTIONAL: Configure
pluginctl.sh
to have the correct paths to Java and your plugin location- Set
PLUGIN_JAVA
to the location of Java on your server (including the "java" filename) - Set
PLUGIN_PATH
to the location of the Unix Plugin
- Set
- Run
./pluginctl.sh start
from command-line - Check logs (in
logs
directory by default) for errors - Login to New Relic UI and find your data in Insights
- In the data explorer, look for custom event types that start with "unixMonitor:"
- Possible event types (for out-of-the-box commands): unixMonitor:Disk, unixMonitor:DiskIO, unixMonitor:NetworkIO, unixMonitor:Process, unixMonitor:Stats, unixMonitor:Vmstat
Note: A full example of the possible fields in plugin.json
can be found in plugin-full-example.json
OS
(default:auto
): Used to determine which commands to run and how to parse them. Leave set toauto
to have the plugin figure that out (which normally works).account_id
: New Relic account ID - the 6- or 7- digit number in the URL when you're logged into the account of your choosing.insights_insert_key
(underinsights
): You must create an Insights Insert key, as described here.insights_data_center
(underinsights
, default:us
): If using the NR EU data center for your account, please change this toeu
orEU
. Otherwise, you can leave this alone or omit this setting entirely. {#eu-data-center}
These settings are found in the agents
object.
name
: If set toauto
, the plugin will use that server's hostname. Otherwise, sets the hostname and agentName to whatever is set here.static
(optional): An object containing static attributes (as name-value pairs) you want to appear in every event from this plugin. For example:
"agents": [
{
"name": "auto",
"static": {
"data_center": "Philadelphia",
"customer": "Eagles",
"rank": 1
}
}
]
If using a proxy, the optional proxy
object should be added to the global
object in plugin.json
, if its not there already.
- The available attributes are:
proxy_host
,proxy_port
,proxy_username
andproxy_password
. - The only attribute that is required in the
proxy
object isproxy_host
.
The plugin can check for and deploy the latest dashboards to your account when it starts up. This requires the dashboards
object in plugin.json
to be set up properly:
admin_api_key
: Admin API key, as described here.integration_guid
: Default isUNIX.Infra.Monitor
. DO NOT DELETE OR CHANGE UNLESS OTHERWISE INSTRUCTED.dashboard_install
: Default iscommand_line
. DO NOT DELETE OR CHANGE UNLESS OTHERWISE INSTRUCTED.- (Optional)
force_deploy
: Default isfalse
, see Redeploying dashboards for details.
If you don't want the dashboard deployment to run at start-up, leave admin_api_key
blank or remove it entirely.
The plugin can force the dashboards to be redeployed by either of two ways:
- Run
./pluginctl.sh dashboards_redeploy
OR - Set
force_deploy
totrue
in thedashboards
object in plugin.json and run./pluginctl.sh dashboards
NOTE: Be sure to re-setforce_deploy
tofalse
once completed. Otherwise, the dashboards will re-deploy every time the./pluginctl.sh start
or./pluginctl.sh restart
is run.
If you want to initiate the dashboard install from a standalone machine (i.e. a tools server or your own mac, linux or cygwin laptop/desktop), you will need the following:
pluginctl.sh
config/plugin.json
(including path) with thedashboard
object filled in as described above.
To install, run ./pluginctl.sh dashboards
.
If you are using the JDK that is packaged with WebSphere and see an exception in the logs like below, it is due to attempting to use the WebSphere SSL Factory instead of the IBM JSSE packages.
ERROR com.newrelic.metrics.publish.binding.Request - An error occurred communicating with the New Relic service
java.net.SocketException: java.lang.ClassNotFoundException: Cannot find the specified class com.ibm.websphere.ssl.protocol.SSLSocketFactory
If so, uncomment the following line in pluginctl.sh
and restart the plugin.
# USE_IBM_JSSE=true
If you see the following error, it may be because the Bourne shell does not support certain syntax in the installer script.
pluginctl.sh: syntax error at line 240: `admin_api_key=$' unexpected
If so, use the Korn shell or Bash (if available). Both were tested successfully in Solaris 10.
If you are trying to customize the commands that the Unix Monitor is running, or you are not seeing any or all of the data you expect to see, you can put the agent into 'Debug Mode', in which it outputs to the logs all of the commands being run and it's attempts at parsing them.
Note: You will need to restart the Unix Monitor to pick up these changes.
This can be enabled EITHER by:
- Running
pluginctl.sh
with thedebug
at the end to start, like so:./pluginctl.sh start debug
or./pluginctl.sh restart debug
OR - Replacing
config/logback.xml
withconfig/logback-debug.xml
and restarting the Unix monitor (if using this method, remember to swap the files back when finished)