good-influx
InfluxDB broadcasting for Good process monitor, based on good-http.
It can write to HTTP or UDP Telegraf endpoints.
Good Influx
will format your Good data according to the InfluxDB Line Protocol.
Usage
good-influx
is a write stream used to send events to InfluxDB endpoints in batches. If your endpoint
is http://
or https://
, it makes a "POST" request with a plain-text payload to the supplied endpoint
. It will make a final "POST" request to the endpoint to flush the rest of the data on "finish".
If the supplied endpoint
is a udp://
endpoint then good-influx
will send the stats via UDP. This may improve application performance since UDP does not wait for a response. Though it does fail silently, so you run the risk that your stats are failing to record and you don't know about it.
Example
const Hapi = require('hapi');
const server = new Hapi.Server();
server.connection();
const options = {
ops: {
interval: 1000
},
reporters: {
influx: [{
module: 'good-squeeze',
name: 'Squeeze',
args: [{ ops: '*' }]
}, {
module: 'good-influx',
args: ['http://localhost:8086/write?db=good', {
threshold: 10,
metadata: {
serviceName: 'SuperAwesomeService',
dataCenter: 'Banff'
},
prefix: ['my', 'awesome', 'service']
}]
}]
}
};
server.register({
register: require('good'),
options: options
}, (err) => {
if (err) {
console.error(err);
} else {
server.start(() => {
console.info('Server started at ' + server.info.uri);
});
}
});
Good Influx
GoodInflux (endpoint, config)
Creates a new GoodInflux object where:
endpoint
- full path to remote server's InfluxDB HTTP API end point to transmit InfluxDB statistics (e.g. http://localhost:8086/write?db=good
)config
- configuration object (Optional)
[threshold]
- number of events to hold before transmission. Defaults to 5
. Set to 0
to have every event start transmission instantly.
- Note that for UDP, threshold above
5
will be set to 5
. Why? Because if UDP packets get too big they fail to transmit.
[errorThreshold]
- number of erroring message sends to tolerate before the plugin fails. Default is 0.[wreck]
- configuration object to pass into wreck
. Defaults to { timeout: 60000, headers: {} }
. content-type
is always "text/plain".[udpType]
- UDP type; defaults to udp4
. Probably not necessary to change, but more documentation is available on the NodeJS Dgram Documentation[metadata]
- arbitrary tags you would like to add to your InfluxDB stats. This helps you query InfluxDB for the statistics you want.[prefix]
- applied to each measurement name. Useful if you want to limit the scope of your measurements to a specific service. You can specify a string, or an array of strings (recommended). Arrays will be joined by prefixDelimiter below. For example, using prefix: ['my', 'awesome', 'service']
the ops
measurement will be renamed to
my/awesome/service/ops
[prefixDelimiter]
- Used to delimit measurement prefix arrays defined in prefix above. Defaults to /
.
Series
Error
time | host | pid | error | id | method | url |
---|
Log
Ops
Each Ops event from the Hapi Good plugin is separated out into 5 events for InfluxDB. Why? Because ops
events are multilayered, so we can't capture the full information in one event.
Standard tags: host,pid, metadata (optional)
event | numEvents | tags | fields |
---|
ops | 1 | Standard | os.cpu1m, os.cpu5m, os.cpu15m, os.freemem, os.totalmem, os.uptime, os.totalmem, proc.delay, proc.heapTotal, proc.heapUsed, proc.rss, proc.uptime |
ops_requests | 1 per port | Standard + port | requestsTotal, requestsDisconnects, requests200* -- one field for each status code |
ops_concurrents | 1 per port | Standard + port | concurrents |
ops_responseTimes | 1 per port | Standard + port | avg, max |
ops_sockets | 1 | Standard | httpTotal, httpsTotal |
Request
time | host | pid | data | id | method | path | tags |
---|
Response
time | host | pid | httpVersion | id | instance | labels | method | path | query |
---|
referer | remoteAddress | responseTime | statusCode | userAgent |
---|
License
MIT