s3-streamlogger

A Writable Stream object that uploads to s3 objects, periodically rotating to a new object name.

See also tails3 for a script to tail the log files produced by s3-streamlogger.

Installation

npm install --save s3-streamlogger

Basic Usage

const { S3StreamLogger } = require('s3-streamlogger');

const s3stream = new S3StreamLogger({
     bucket: "mys3bucket",
});

s3stream.write("hello S3");

Use with Winston: Log to S3

npm install --save winston
npm install --save s3-streamlogger

const winston            = require('winston');
const { S3StreamLogger } = require('s3-streamlogger');

const s3_stream = new S3StreamLogger({
     bucket: "mys3bucket",
});

const transport = new (winston.transports.Stream)({
  stream: s3_stream
});
// see error handling section below
transport.on('error', function(err){/* ... */});

const logger = winston.createLogger({
  transports: [transport]
});

logger.info('Hello Winston!');

Define subfolder

const { S3StreamLogger } = require('s3-streamlogger');

const s3stream = new S3StreamLogger({
     bucket: "mys3bucket",
     folder: "my/nested/subfolder"
});

s3stream.write("hello S3");

Assign tags

const { S3StreamLogger } = require('s3-streamlogger');

const s3stream = new S3StreamLogger({
     bucket: "mys3bucket",
     folder: "my/nested/subfolder",
       tags: {type: 'myType', project: 'myProject'}
});

s3stream.write("hello S3");

Add hostname information for tails3

tails3 expects messages to be logged as json (the default for the file transport), with hostname and (for critical errors), stack properties to each log object, in addition to the standard timestamp, level and message properties. You can provide these using the third "metadata" option to winston's log method:

logger.log(level, message, {hostname: ... , stack: ...});

Handling logging errors

When there is an error writing to s3, the stream emits an 'error' event with details. You should take care not to log these errors back to the same stream (as that is likely to cause infinite recursion). Instead log them to the console, to a file, or to SNS using winston-sns.

Note that these errors will result in uncaught exceptions unless you have an error event handler registered, for example:

s3_stream.on('error', function(err){
    // there was an error!
    some_other_logging_transport.log('error', 'logging transport error', err)
});

When using s3-streamlogger with the Winston Stream transport, the Stream transport attaches its own error handler to the stream, so you do not need your own, however it will re-emit the errors on itself which must be handled instead:

const transport = new (winston.transports.Stream)({
  stream: s3_stream
});
transport.on('error', function(err){
  /* handle s3 stream errors (e.g. invalid credentials, EHOSTDOWN) here */
});

const logger = winston.createLogger({
  transports: [transport]
});

Options

bucket (required)

Name of the S3 bucket to upload data to. Must exist. Can also be provided as the environment variable BUCKET_NAME.

folder

An optional folder to stream log files to. Takes a path string, eg: "my/subfolder" or "nested".

tags

An optional set of tags to assign to the log files. Takes an object, eg: {type: "myType"} or {type: "myType", project: "myProject"}.

access_key_id deprecated

AWS access key ID, must have putObject permission on the specified bucket. Provide credentials through the environment variable AWS_ACCESS_KEY_ID, or as any of the other authentication methods supported by the AWS SDK instead.

secret_access_key deprecated

AWS secret key for the access_key_id specified. Provide credentials through the environment variable AWS_SECRET_ACCESS_KEY, or as any of the other authentication methods supported by the AWS SDK instead.

config

Configuration object for the AWS SDK. The full list of options is available on the AWS SDK Configuration page. This is an alternative to using access_key_id and secret_access_key and is overwritten by them if both are used.

name_format

Format of file names to create, accepts strftime specifiers. Defaults to "%Y-%m-%d-%H-%M-%S-%L-unknown-<hostname>.log". The Date() used to fill the format specifiers is created with the current UTC time, but still has the current timezone, so any specifiers that perform timezone conversion will return incorrect dates.

If you use a format of the form %Y-%m-%d-%H-%M-<stage>-<hostname>.log, then you can use tails3 to tail the log files being generated by S3StreamLogger.

If compress is set to true, then the default extension is .log.gz instead of .log.

rotate_every

Files will be rotated every rotate_every milliseconds. Defaults to 3600000 (60 minutes).

max_file_size

Files will be rotated when they reach max_file_size bytes. Defaults to 200000 (i.e. 200 KB).

upload_every

Files will be uploaded every upload_every milliseconds. Defaults to 20 seconds.

buffer_size

Files will be uploaded if the un-uploaded data exceeds buffer_size bytes. Defaults to 10 KB.

server_side_encryption

The server side encryption AES256 algorithm used when storing objects in S3. Defaults to false.

storage_class

The S3 StorageClass (STANDARD, REDUCED_REDUNDANCY, etc.). If omitted, no value is used and aws-sdk will fill in its default.

acl

The canned ACL (access control list) to apply to uploaded objects. Defaults to no ACL.

compress

If true, the files will be gzipped before uploading (may reduce s3 storage costs). Defaults to false.

License

ISC: equivalent to 2-clause BSD.