This library contains functions to enable distributed tracing & logging in University of Auckland AWS Lambda projects.
If deploying to Kong gateway, your lambda plugin should have the awsgateway_compatible
flag set to true
.
This ensures the trace headers will be passed through to your lambda project correctly, but you will now have to base64 decode the request body before using it.
Install the library using the command
npm install @uoa/lambda-tracing
In your src folder, create a new file tracing-wrapper.ts
In this, add the following code
import {initializeTracing} from "@uoa/lambda-tracing";
initializeTracing();
In your lambda environment.yml file, add the following NODE_OPTIONS
environment variable in the properties map
lambda:
properties:
NODE_OPTIONS: --require src/tracing-wrapper
The above ensures that the tracing-wrapper.ts
code will be loaded before your lambda app starts, which will
perform the required setup for logging and distributed tracing.
Whenever you want to use the logging provided in this library, simply import the logger using
const logger = require('@uoa/lambda-tracing/logging')(module);
Note: (module)
must be passed as a parameter to this import so that the logger knows where it has been called from.
Once the logger has been imported, you can log any info using
logger.info('Hello Logger!');
Logging Levels
At UoA, we have five logging levels available to use. Ordered by decreasing priority, these are:
Level | Logger usage |
---|---|
ERROR | logger.error() |
WARN | logger.warn() |
AUDIT | logger.audit() |
INFO | logger.info() |
DEBUG | logger.debug() |
By default, any log at the INFO
level or higher will be logged. However, this can be changed by adding another
environment variable loggingLevel
in the lambda properties map of the environment.yml file. The value of this
specifies the lowest level that will be logged.
lambda:
properties:
NODE_OPTIONS: --require src/tracing-wrapper
loggingLevel: WARN
logger.debug('Will not be logged');
logger.info('Will not be logged');
logger.audit({...}, 'Will not be logged');
logger.warn('Will be logged');
logger.error('Will be logged');
Logging format
By default, logs will be produced in the following format:
date [thread] level class - [[[traceId,spanId,info]]] message
This can also be changed by adding another environment variable loggingPattern
in the lambda properties map
of the environment.yml file.
lambda:
properties:
NODE_OPTIONS: --require src/tracing-wrapper
loggingPattern: "%date - %message | %level | %class"
Valid logging pattern placeholders are as follows:
Placeholder | Replacement value | Example value |
---|---|---|
%date | Date time when the message is logged in ISO8601 format | 2022-05-29T21:31:11.250Z |
%thread | Unused, this is only present in the default pattern to match with UoA logging standards. Value will always be -
|
- |
%level | Log level | INFO |
%class | The module which logged the message | src.handle-service |
%traceId | Unique Id for the request | 5c0c783c965a684842608f10422fdf2c |
%spanId | Unique Id for the current lambda invocation | 6a8b025511ac1191 |
%info | Extra information to help with filtering the logs | usefulCode=12345 |
%message | The logged message | Hello Logger! |
Audit Logs
All logger functions except logger.audit()
take a single string as the message to be logged. The audit() function
takes an AuditInformation
object as a parameter. A second message
parameter of type string is optional. Using the
values passed in the AuditInformation
parameter, the audit log message will be standardised so that logstash can
correctly ingest and send the logs to the uoa-app-audit-logs-* index on Kibana.
The AuditInformation
parameter has the below required properties:
Property | Type | Description |
---|---|---|
application (optional) | string | Application name that groups different physical systems into a single logical group. For example, this may be the front end system name so that multiple backend system logs can be grouped together. |
action | 'read' | 'update' | 'create' | 'delete' | The CRUD action being performed on the resource. |
resourceId (optional) | string | The ID of the resource being actioned upon. |
resourceType | string | The name/type of the resource being actioned upon. |
ownerId | string | The ID of the owner of the resource information. |
accessedBy | string | The ID of the person/system accessing the resource. |
Calling the logger.audit()
function as below:
logger.audit({application: "applicationName", action: "read", resourceId: "123456789", resourceType: "personalDetails", accessedBy: "upi1234", ownerId: "upi5678"}, "Additional useful info");
Will produce the following log message:
2024-04-12T00:58:05.321Z [-] AUDIT src.controller.person-controller - [[[TraceId,SpanId,Info]]] application:applicationName accessedBy:upi1234 action:read owner:upi5678 resourceType:personalDetails resourceId:123456789 Additional useful info
There are two headers that can be passed in API calls to your lambda project using this library: X-B3-TraceId
and X-B3-Info
.
If passed to a lambda project using this library, the values of %traceId and %info in the logging pattern will be the corresponding header values.
X-B3-TraceId
should be either a 16 or 32 character lowercase hex encoded string.
If the passed traceId is less than 16 characters, it will be left padded with 0's to get to a length of 16.
Likewise, if it is longer than 16 but less than 32 characters, it will be left padded with 0's to get to a length of 32.
If it is longer than 32 characters, a new traceId will be generated.
If the X-B3-TraceId
and X-B3-SpanId
headers are not passed, they will be randomly generated.
X-B3-Info
can be any string value to provide extra information in your logs.
The value of this header can be accessed by using the getTraceInfoHeader(): string
function.
Similarly, it can be set using the setTraceInfoHeader(string)
function.
These two functions can be imported using:
const {setTraceInfoHeader, getTraceInfoHeader} = require('./logging/tracing');
Header Propagation
To propagate the X-B3-TraceId
and X-B3-Info
headers to other APIs, you can import and use the uoaHttps module:
const uoaHttps = require('@uoa/lambda-tracing/uoaHttps');
The uoaHttps
module exposes some functions to perform the primary HTTP operations. These will inject the tracing headers into requests before they are made:
doGetRequest(hostname, path, headers): Promise
doPostRequest(hostname, path, headers, data): Promise
doPutRequest(hostname, path, headers, data): Promise
doDeleteRequest(hostname, path, headers): Promise
There is also another function request()
exposed in this module in case header propagation with operations other than the basic GET, POST, PUT, and DELETE are required.
The usage of this is the same as the one provided by the Node https library (see specs here).