SFTPGo Pub/Sub plugin

This plugin allows to send SFTPGo filesystem and provider events to publish/subscribe systems. It is not meant to react to pre-* events. It simply forwards the configured events to an external pub/sub system.

Configuration

The supported services can be configured within the plugins section of the SFTPGo configuration file: you have to set the service publish URL as first plugin argument. You can also set an, optional, instance id by passing a second argument. The instance id is useful if you are receiving notifications from multiple SFTPGo instances and want to identify where the events are coming from.

This is an example configuration.

...
"plugins": [
    {
      "type": "notifier",
      "notifier_options": {
        "fs_events": [
          "download",
          "upload"
        ],
        "provider_events": [
          "add",
          "delete"
        ],
        "provider_objects": [
          "user",
          "admin",
          "api_key"
        ],
        "retry_max_time": 60,
        "retry_queue_max_size": 1000
      },
      "cmd": "<path to sftpgo-plugin-pubsub>",
      "args": ["rabbit://sftpgo"],
      "sha256sum": "",
      "auto_mtls": true
    }
  ]
...

With the above example the plugin is configured to connect to RabbitMQ and publish messages to the sftpgo fanout exchange. The RabbitMQ’s server is discovered from the RABBIT_SERVER_URL environment variable (which is something like amqp://guest:guest@localhost:5672/).

The plugin will not start if it fails to connect to the configured service, this will prevent SFTPGo from starting.

The plugin will panic if an error occurs while publishing an event, for example because the service connection is lost. SFTPGo will automatically restart it and you can configure SFTPGo to retry failed events until they are older than a configurable time (60 seconds in the above example). This way no event is lost.

Notifications format

The filesystem events will contain a JSON serialized struct in the message body with the following fields:

timestamp, string formatted as RFC3339 with nanoseconds precision
action, string, an SFTPGo supported action
username
fs_path, string filesystem path
fs_target_path, string, included for rename action and sftpgo-copy SSH command
virtual_path, string, path seen by SFTPGo users
virtual_target_path, string, target path seen by SFTPGo users
ssh_cmd, string, included for ssh_cmd action
file_size, integer, included for pre-upload, upload, download, delete actions if the file size is greater than 0
elapsed, integer. Elapsed time as milliseconds
status, integer. 1 means no error, 2 means a generic error occurred, 3 means quota exceeded error
protocol, string. Possible values are SSH, SFTP, SCP, FTP, DAV, HTTP, HTTPShare, DataRetention, OIDC, EventAction
ip, string. The action was executed from this IP address
session_id, string. Unique protocol session identifier. For stateless protocols such as HTTP the session id will change for each request
fs_provider, int. Filesystem provider 0 for local filesystem, 1 for S3 backend, 2 for Google Cloud Storage (GCS) backend, 3 for Azure Blob Storage backend, 4 for local encrypted backend, 5 for SFTP backend
bucket, string. Non-empty for S3, GCS and Azure backends
endpoint, string. Non-empty for S3, SFTP and Azure backend if configured
open_flags, integer. File open flags, can be non-zero for pre-upload action. If file_size is greater than zero and open_flags&512 == 0 the target file will not be truncated
role, string. Included if the user who executed the action has a role
instance_id, string. Included if you pass an instance id as the second CLI parameter

The action is also added as metadata.

The provider events will contain a JSON serialized struct in the message body with the following fields:

timestamp, string formatted as RFC3339 with nanoseconds precision
action, string, an SFTPGo supported action
username, string, the username that executed the action. There are two special usernames: __self__ identifies a user/admin that updates itself and __system__ identifies an action that does not have an explicit executor associated with it, for example users/admins can be added/updated by loading them from initial data
ip, string. The action was executed from this IP address
object_type, string. Afftected object type, for example user, admin, api_key
object_name, string. Unique identifier for the affected object, for example username or key id
object_data, base64 of the JSON serialized object with sensitive fields removed
role, string. Included if the admin/user who executed the action has a role
instance_id, string. Included if you pass an instance id as the second CLI parameter

The action and the object_typeare also added as metadata.

Supported services

We use Go CDK to access several publish/subscribe systems in a portable way.

Google Cloud Pub/Sub

To publish events to a Google Cloud Pub/Sub topic, you have to use gcppubsub as URL scheme and you have to include the project ID and the topic ID within the URL, for example gcppubsub://projects/myproject/topics/mytopic.

This plugin will use Application Default Credentials. See here for alternatives such as environment variables.

Amazon Simple Notification Service

To publish events to an Amazon Simple Notification Service (SNS) topic, you have to use awssns as URL scheme. The topic is identified via the Amazon Resource Name (ARN). You should specify the region query parameter to ensure your application connects to the correct region.

Here is a sample URL: awssns:///arn:aws:sns:us-east-2:123456789012:mytopic?region=us-east-2.

The plugin will create a default AWS Session with the SharedConfigEnable option enabled. See AWS Session to learn about authentication alternatives, including using environment variables.

Amazon Simple Queue Service

To publish events to an Amazon Simple Queue Service (SQS) topic, you have to use an URL that closely resemble the queue URL, except the leading https:// is replaced with awssqs://. You can specify the region query parameter to ensure your application connects to the correct region.

Here is a sample URL: awssqs://sqs.us-east-2.amazonaws.com/123456789012/myqueue?region=us-east-2.

Azure Service Bus

To publish to an Azure Service Bus topic over AMQP 1.0, you have to use azuresb as URL scheme and the topic name as URL host.

Here is a sample URL: azuresb://mytopic.

We use the environment variable SERVICEBUS_CONNECTION_STRING to obtain the Service Bus connection string. The connection string can be obtained from the Azure portal.

RabbitMQ

To publish events to an AMQP 0.9.1 fanout exchange, the dialect of AMQP spoken by RabbitMQ, you have to use rabbit as URL scheme and the exchange name as URL host.

Here is a sample URL: rabbit://myexchange.

The RabbitMQ’s server is discovered from the RABBIT_SERVER_URL environment variable (which is something like amqp://guest:guest@localhost:5672/).

NATS

To publish events to a NATS subject, you have to use nats as URL scheme and the subject name as URL host.

Here is a sample URL: nats://example.mysubject.

The NATS server is discovered from the NATS_SERVER_URL environment variable (which is something like nats://nats.example.com).

Kafka

To publish events to a Kafka cluster, you have to use kafka as URL scheme and the topic name as URL host.

Here is a sample URL: kafka://my-topic.

The brokers in the Kafka cluster are discovered from the KAFKA_BROKERS environment variable (which is a comma-delimited list of hosts, something like 1.2.3.4:9092,5.6.7.8:9092).

joefrancia/sftpgo-plugin-pubsub