Skip to content

Latest commit

 

History

History
451 lines (327 loc) · 16.2 KB

index.asciidoc

File metadata and controls

451 lines (327 loc) · 16.2 KB

S3 output plugin

Description

This plugin batches and uploads logstash events into Amazon Simple Storage Service (Amazon S3).

Important
The S3 output plugin only supports AWS S3. Other S3 compatible storage solutions are not supported.

S3 outputs create temporary files into the OS' temporary directory. You can specify where to save them using the temporary_directory option.

Important
For configurations containing multiple s3 outputs with the restore option enabled, each output should define its own temporary_directory. Shared or nested directories can cause data loss upon recovery.
Requirements
  • Amazon S3 Bucket and S3 Access Permissions (Typically access_key_id and secret_access_key)

  • S3 PutObject permission

S3 output file
`ls.s3.312bc026-2f5d-49bc-ae9f-5940cf4ad9a6.2013-04-18T10.00.tag_hello.part0.txt`

ls.s3

indicates logstash plugin s3

312bc026-2f5d-49bc-ae9f-5940cf4ad9a6

a new, random uuid per file.

2013-04-18T10.00

represents the time whenever you specify time_file.

tag_hello

indicates the event’s tag.

part0

If you indicate size_file, it will generate more parts if your file.size > size_file. When a file is full, it gets pushed to the bucket and then deleted from the temporary directory. If a file is empty, it is simply deleted. Empty files will not be pushed.

Crash Recovery

This plugin will recover and upload temporary log files after crash/abnormal termination when using restore set to true

Usage

This is an example of logstash config:

output {
   s3{
     access_key_id => "crazy_key"             (optional)
     secret_access_key => "monkey_access_key" (optional)
     region => "eu-west-1"                    (optional, default = "us-east-1")
     bucket => "your_bucket"                  (required)
     size_file => 2048                        (optional) - Bytes
     time_file => 5                           (optional) - Minutes
     codec => "plain"                         (optional)
     canned_acl => "private"                  (optional. Options are "private", "public-read", "public-read-write", "authenticated-read", "aws-exec-read", "bucket-owner-read", "bucket-owner-full-control", "log-delivery-write". Defaults to "private" )
   }

S3 Output Configuration Options

This plugin supports the following configuration options plus the [plugins-outputs-s3-common-options] described later.

Setting Input type Required

access_key_id

string

No

additional_settings

hash

No

aws_credentials_file

string

No

bucket

string

Yes

canned_acl

string, one of ["private", "public-read", "public-read-write", "authenticated-read", "aws-exec-read", "bucket-owner-read", "bucket-owner-full-control", "log-delivery-write"]

No

encoding

string, one of ["none", "gzip"]

No

endpoint

string

No

prefix

string

No

proxy_uri

string

No

region

string

No

restore

boolean

No

retry_count

number

No

retry_delay

number

No

role_arn

string

No

role_session_name

string

No

rotation_strategy

string, one of ["size_and_time", "size", "time"]

No

secret_access_key

string

No

server_side_encryption

boolean

No

server_side_encryption_algorithm

string, one of ["AES256", "aws:kms"]

No

session_token

string

No

signature_version

string, one of ["v2", "v4"]

No

size_file

number

No

ssekms_key_id

string

No

storage_class

string, one of ["STANDARD", "REDUCED_REDUNDANCY", "STANDARD_IA", "ONEZONE_IA"]

No

temporary_directory

string

No

time_file

number

No

upload_multipart_threshold

number

No

upload_queue_size

number

No

upload_workers_count

number

No

validate_credentials_on_root_bucket

boolean

No

Also see [plugins-outputs-s3-common-options] for a list of options supported by all output plugins.

 

access_key_id
  • Value type is string

  • There is no default value for this setting.

This plugin uses the AWS SDK and supports several ways to get credentials, which will be tried in this order:

  1. Static configuration, using access_key_id and secret_access_key params in logstash plugin config

  2. External credentials file specified by aws_credentials_file

  3. Environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY

  4. Environment variables AMAZON_ACCESS_KEY_ID and AMAZON_SECRET_ACCESS_KEY

  5. IAM Instance Profile (available when running inside EC2)

additional_settings
  • Value type is hash

  • Default value is {}

Key-value pairs of settings and corresponding values used to parametrize the connection to S3. See full list in the AWS SDK documentation. Example:

    output {
      s3 {
        access_key_id => "1234",
        secret_access_key => "secret",
        region => "eu-west-1",
        bucket => "logstash-test",
        additional_settings => {
          "force_path_style" => true,
          "follow_redirects" => false
        }
      }
    }
aws_credentials_file
  • Value type is string

  • There is no default value for this setting.

Path to YAML file containing a hash of AWS credentials. This file will only be loaded if access_key_id and secret_access_key aren’t set. The contents of the file should look like this:

    :access_key_id: "12345"
    :secret_access_key: "54321"
bucket
  • This is a required setting.

  • Value type is string

  • There is no default value for this setting.

S3 bucket

canned_acl
  • Value can be any of: private, public-read, public-read-write, authenticated-read, aws-exec-read, bucket-owner-read, bucket-owner-full-control, log-delivery-write

  • Default value is "private"

The S3 canned ACL to use when putting the file. Defaults to "private".

encoding
  • Value can be any of: none, gzip

  • Default value is "none"

Specify the content encoding. Supports ("gzip"). Defaults to "none"

endpoint
  • Value type is string

  • There is no default value for this setting.

The endpoint to connect to. By default it is constructed using the value of region. This is useful when connecting to S3 compatible services, but beware that these aren’t guaranteed to work correctly with the AWS SDK. The endpoint should be an HTTP or HTTPS URL, e.g. https://example.com

prefix
  • Value type is string

  • Default value is ""

Specify a prefix to the uploaded filename to simulate directories on S3. Prefix does not require leading slash. This option supports {logstash-ref}/event-dependent-configuration.html#sprintf[Logstash interpolation]. For example, files can be prefixed with the event date using prefix = "%{+YYYY}/%{+MM}/%{+dd}".

Important
Take care when you are using interpolated strings in prefixes. This has the potential to create large numbers of unique prefixes, causing large numbers of in-progress uploads. This scenario may result in performance and stability issues, which can be further exacerbated when you use a rotation_strategy that delays uploads.
proxy_uri
  • Value type is string

  • There is no default value for this setting.

URI to proxy server if required

region
  • Value type is string

  • Default value is "us-east-1"

The AWS Region

restore
  • Value type is boolean

  • Default value is true

Used to enable recovery after crash/abnormal termination. Temporary log files will be recovered and uploaded.

Note
If you’re using multiple S3 outputs, always set temporary_directory to a unique directory. Otherwise the recovery mechanism won’t work correctly.
retry_count
  • Value type is number

  • Default value is Infinity

Allows to limit number of retries when S3 uploading fails.

retry_delay
  • Value type is number

  • Default value is 1

Delay (in seconds) to wait between consecutive retries on upload failures.

role_arn
  • Value type is string

  • There is no default value for this setting.

The AWS IAM Role to assume, if any. This is used to generate temporary credentials, typically for cross-account access. See the AssumeRole API documentation for more information.

role_session_name
  • Value type is string

  • Default value is "logstash"

Session name to use when assuming an IAM role.

rotation_strategy
  • Value can be any of: size_and_time, size, time

  • Default value is "size_and_time"

Controls when to close the file and push it to S3.

If you set this value to size, it uses the value set in size_file. If you set this value to time, it uses the value set in time_file. If you set this value to size_and_time, it uses the values from size_file and time_file, and splits the file when either one matches.

The default strategy checks both size and time. The first value to match triggers file rotation.

secret_access_key
  • Value type is string

  • There is no default value for this setting.

The AWS Secret Access Key

server_side_encryption
  • Value type is boolean

  • Default value is false

Specifies whether or not to use S3’s server side encryption. Defaults to no encryption.

server_side_encryption_algorithm
  • Value can be any of: AES256, aws:kms

  • Default value is "AES256"

Specifies what type of encryption to use when SSE is enabled.

session_token
  • Value type is string

  • There is no default value for this setting.

The AWS Session token for temporary credential

signature_version
  • Value can be any of: v2, v4

  • There is no default value for this setting.

The version of the S3 signature hash to use. Normally uses the internal client default, can be explicitly specified here

size_file
  • Value type is number

  • Default value is 5242880

Set the file size in bytes. When the number of bytes exceeds the size_file value, a new file is created. If you use tags, Logstash generates a specific size file for every tag.

ssekms_key_id
  • Value type is string

  • There is no default value for this setting.

The key to use when specified along with server_side_encryption ⇒ aws:kms. If server_side_encryption ⇒ aws:kms is set but this is not default KMS key is used. http://docs.aws.amazon.com/AmazonS3/latest/dev/UsingKMSEncryption.html

storage_class
  • Value can be any of: STANDARD, REDUCED_REDUNDANCY, STANDARD_IA, ONEZONE_IA

  • Default value is "STANDARD"

Specifies what S3 storage class to use when uploading the file. More information about the different storage classes can be found: http://docs.aws.amazon.com/AmazonS3/latest/dev/storage-class-intro.html Defaults to STANDARD.

temporary_directory
  • Value type is string

  • Default value is "/tmp/logstash"

Set the directory where logstash will store the tmp files before sending it to S3 default to the current OS temporary directory in linux /tmp/logstash.

Warning
Using multiple S3 outputs with restore ⇒ true requires unique directories per output. All of the directory’s contents are processed and deleted upon recovery, and shared or nested directories can cause data loss. For example, an output using /tmp/s3 and a second configured with /tmp/s3/sub would cause issues. Having temporary directories /tmp/s3/sub1 and /tmp/s3/sub2 is fine.
time_file
  • Value type is number

  • Default value is 15

Set the time, in MINUTES, to close the current sub_time_section of bucket. If rotation_strategy is set to time or size_and_time, then time_file cannot be set to 0. Otherwise, the plugin raises a configuration error.

upload_multipart_threshold
  • Value type is number

  • Default value is 15728640

Files larger than this number are uploaded using the S3 multipart APIs

upload_queue_size
  • Value type is number

  • Default value is 4

Number of items we can keep in the local queue before uploading them

upload_workers_count
  • Value type is number

  • Default value is 4

Specify how many workers to use to upload the files to S3

validate_credentials_on_root_bucket
  • Value type is boolean

  • Default value is true

The common use case is to define permissions on the root bucket and give Logstash full access to write logs. In some circumstances, you need more granular permissions on the subfolder. This allows you to disable the check at startup.