@iopipe/trace
v1.6.1
Published
IOpipe plugin for tracing metrics
Downloads
398
Readme
IOpipe Trace Plugin
Create marks and measures for arbitrary units of time. Measure latency of database calls, third party requests, or code blocks and visualize them in IOpipe!
Supports automatic tracing of functions using Node's native http
and https
, as well as ioredis
.
Requirements
- Node >=
8.10.0
- NPM >=
6.9.0
- IOpipe >=
1.x
Install
Note: This plugin is automatically included in the recommended package @iopipe/iopipe
With yarn (recommended) in project directory:
yarn add @iopipe/trace
With npm in project directory:
npm install @iopipe/trace
Then include the plugin with IOpipe in your serverless function:
const iopipeLib = require('@iopipe/iopipe');
const tracePlugin = require('@iopipe/trace');
const iopipe = iopipeLib({
token: 'TOKEN_HERE',
plugins: [tracePlugin()]
});
// wrap your lambda handler
exports.handler = iopipe((event, context) => {
const {mark} = context.iopipe;
mark.start('database');
// after database call is finished
mark.end('database');
mark.start('analytics');
// after analytics call is finished
mark.end('analytics');
context.succeed('Wow!');
});
Methods
// create the start mark
// the string argument is a name you are assigning the particular trace
context.iopipe.mark.start('db');
// create the end mark
// pass the name of the trace that you want to end
context.iopipe.mark.end('db');
// create an custom measurement between start:init and end:db
context.iopipe.measure('custom', 'init', 'db');
Config
autoHttp
(object: optional = {})
Automatically create traces and matching metadata for each http/s call made within your function invocation.
autoHttp.enabled
(bool: optional = false)
Enable HTTP/S auto-tracing. Enabled by default. You can also use the environment variable IOPIPE_TRACE_AUTO_HTTP_ENABLED
.
autoHttp.filter
(func: optional)
Filter what data is recorded for each http request that occurs within the invocation. The function will be passed two arguments. The first is an object containing "safe" data that is typically not sensitive in nature. The second argument is a more complete object with the same shape that may include sensitive data. You can use these objects to determine:
- A. What information to record / not to record (i.e. filter out certain headers)
- B. If the http call should be completely excluded from trace data (ie filter out sensitive calls altogether)
const iopipe = iopipeLib({
plugins: [
tracePlugin({
autoHttp: {
enabled: true,
filter: (safeData, allData) => {
// obj = {'request.url':'http://iopipe.com', 'request.method': 'GET'}
// return the object with filtered keys
// or return false to exclude the trace data completely
const url = safeData['request.url'];
if (url.match(/restricted/)) {
// if you don't want any traces on this restricted URI return false
return false;
} else if (url.match(/cat-castle/)) {
// if you need to keep track of a sensitive header
return Object.assign(safeData, {
'request.headers.cookie': allData['request.headers.cookie']
});
}
// if you want to record the default data after some logic checks
return safeData;
}
}
})
]
});
autoRedis
Automatically trace Redis commands using redis (node_redis)
Set the environment variable IOPIPE_TRACE_REDIS
to true
, and IOpipe will trace Redis commands automatically: which command, which key is being read or written, and information about the connection: hostname, port, and connection name (if defined in your connection options). Commands batched with multi/exec are traced individually, so you can measure individual performance within batch operations.
If you're using [email protected] or earlier, turn on auto-tracing with the IOPIPE_TRACE_REDIS_CB
environment variable set to true.
autoIoRedis
Automatically trace Redis commands using ioredis
Set the environment variable IOPIPE_TRACE_IOREDIS
to true
, and your function will enable automatic traces on Redis commands: the name of the command, name of the host, port, and connection (if defined in your connection options), and the key being written or read. Commands batched with multi/exec are traced individually, so you can measure individual performance within batch operations.
autoMongoDb
Automatically trace MongoDB commands
Set the environment variable IOPIPE_TRACE_MONGODB
to true
, and IOpipe will trace commands automatically: which command, which key is being read or written, database and collection name (if available), and the connection's hostname and port. This plugin supports these commands:
command
,insert
,update
, andremove
on the Server classconnect
,close
, anddb
on the MongoClient class.find
,findOne
,insertOne
,insertMany
,updateOne
,updateMany
,replaceOne
,deleteOne
,deleteMany
,createIndex
on collections.bulkWrite
is also supported, and generates a list of which commands were part of the bulk operation.next
,filter
,sort
,hint
, andtoArray
on the Cursor class.
Commands used with a callback parameter generate end traces and duration metrics. (Note that MongoDB commands that don't take callback params--like find
--won't generate durations.)
This plugin supports MongoDB Node.JS Driver v3.3 and newer.
autoMeasure
(bool: optional = true)
By default, the plugin will create auto-measurements for marks with matching mark.start
and mark.end
. These measurements will be displayed in the IOpipe Dashboard. If you'd like to turn this off, set autoMeasure: false
.
const iopipe = iopipeLib({
plugins: [tracePlugin({
autoMeasure: false
})]
});
Contributing
- This project uses Prettier. Please execute
npm run eslint -- --fix
to auto-format the code before submitting pull requests.