This project is based on Pino Glef
GELF transformer is a tool which receives json formatted logs from the stdin and transforms them into GELF format GELF It can also use a custom mapping schema to fill the output log with more data.
npm i -g gelf-transformer
If your application is pushing logs to the standard output then pipe them to gelf transformer.
node your-app.js | gelf-transformer log <options>
Switch | Description | Default | Notes |
---|---|---|---|
-h |
Host | 127.0.0.1 |
Graylog server host |
-p |
Port | 12201 |
Graylog server port |
-m |
Maximum Chunk Size | 1420 |
|
-c |
Custom schema | false |
You can provide a schema which will define which information from your original logs will be visible in the graylog formatted log |
-v |
Verbose mode | false |
Output GELF to console |
-t |
Start sending logs to Graylog | false |
It will start to send logs to the defined graylog server |
Given the log message (formatted as JSON for readability):
{
"pid":16699,
"hostname":"han",
"name":"gelf-test-app",
"level":30,
"time":1481840140708,
"msg":"request completed",
"customField":"test",
"res":{"statusCode":304},
"responseTime":8,
"req":{
"method":"GET",
"headers":{
"host":"localhost:3000",
"user-agent":"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_1) AppleWebKit/602.2.14 (KHTML, like Gecko) Version/10.0.1 Safari/602.2.14"}
},
"v":1
}
Given custom schema json file (my_custom_schema.json):
{
"title": "GELF Schema",
"type": "object",
"properties": {
"_status_code": {
"type": "integer",
"source": "res.statusCode"
},
"_user_agent": {
"type": "string",
"source": "req.headers.user-agent"
},
"customField": {
"type": "string"
}
}
}
And the usage:
node server.js | gelf-transformer log -v -c my_custom_schema.json
Gelf Transformer will show the following message to your Graylog server (formatted here as JSON for readability):
{
"version":"1.1",
"host":"han",
"short_message":"request completed",
"full_message":"request completed",
"timestamp":1481840140.708,
"level":6,
"facility":"gelf-test-app",
"_status_code":304,
"_user_agent":"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_1) AppleWebKit/602.2.14 (KHTML, like Gecko) Version/10.0.1 Safari/602.2.14",
"customField":"test"
}
Right now automatic mapping of fields is done as follows:
Output GELF | Input log | Notes |
---|---|---|
version |
- |
Hardcoded to 1.1 per GELF docs |
host |
hostname |
|
short_message |
msg |
This message is truncated to 64 characters |
full_message |
msg |
msg is not truncated |
timestamp |
time |
|
level |
level |
Default level codes from Pino are mapped to SysLog levels1 |
facility |
name |
deprecated |
By default Gelf Transfomer will log level from a Pino format to syslog format:
Pino Log Level Value | Pino Log Level Name | SysLog Level |
---|---|---|
10 | Trace | Debug |
20 | Debug | Debug |
30 | Info | Info |
40 | Warn | Warning |
50 | Error | Error |
60 | Fatal | Critical |
Note: A log messages without a level map to SysLog Critical
TBD
The implementation of Pino GELF is based in large part on pino-syslog and gelf-node.