Skip to content

Latest commit

 

History

History
338 lines (256 loc) · 29.6 KB

File metadata and controls

338 lines (256 loc) · 29.6 KB

SETUP

The ATT&CK Workbench TAXII server is a Node.js server designed to serve STIX 2.1 content through a TAXII 2.1 compliant REST API. It is loosely coupled to ATT&CK Workbench in that users are empowered to connect their own STIX repositories. By default, the TAXII server sources all STIX data from the Workbench REST API. The TAXII server can run in several deployment models:

  1. As a standalone Node.js instance.
  2. As a standalone Docker container.
  3. As a multi-container Workbench ensemble via Docker Compose.

Endpoints

All available HTTP endpoints are summarized in the taxii module here.

While the TAXII protocol does allow for POST requests, this implementation is read-only and does not provide the means to write data to the local knowledge base.

There are a few notable endpoint-related deviations from the TAXII 2.1 specifications:

  1. The POST ("Add an Object") and DELETE ("Delete An Object") endpoints are not implemented. The reason for these exclusions come from the fundamental design decision to loosely couple the TAXII 2.1 server to Workbench. The TAXII 2.1 server reads STIX data from the Workbench REST API, and re-expresses the data in accordance with the TAXII specification. In other words, data flow is unidirectional. We opted to omit any TAXII endpoints that duplicate existing Workbench functionality, which happens to include the aforementioned POST and DELETE endpoints. However, we recognize that some community members may prefer to not use Workbench, so we designed the stix module to be capable of loading data from other sources. More details on the stix module can be found here.
  2. The "Get Status" endpoint is not implemented. This endpoint only serves to monitor the status of POST requests (i.e., adding objects to a collection). This endpoint was omitted because there are no POST endpoints to monitor.

Environment Variables

All environment variables are assessed at runtime. Any subsequent changes (after the server is started) will require the server to be rebooted for the changes to take effect.

At present, the following environment variables are supported:

Name Type Default Value Description
TAXII_APP_ADDRESS str 0.0.0.0 Specifies which network interface the server should bind to
TAXII_APP_PORT int 443 Specifies which port the server should listen on
TAXII_MAX_CONTENT_LENGTH int 0 Limits the maximum size of the message body that the server will accept
TAXII_API_ROOT_PATH str /api/v21/ Prefix for all TAXII 2.1 endpoints (excluding the Discovery endpoint [/taxii2/])
TAXII_API_ROOT_TITLE str MITRE ATT&CK TAXII 2.1 A title for the API root (observed on responses to the 'Get API Root Information' endpoint)
TAXII_API_ROOT_DESCRIPTION str This API Root contains TAXII 2.1 REST API endpoints that serve MITRE ATT&CK STIX 2.1 data A summary or description of the API root
TAXII_CONTACT_EMAIL str [email protected] The email address which is advertised on responses to the Discovery endpoint (/taxii2/)
TAXII_CACHE_TYPE str default Specifies what type of cache the TAXII server should use; Supported values: 'default' or 'memcached'
TAXII_CACHE_HOST str localhost IP or FQDN of the cache host. Supports multiple comma-separated hosts. e.g., a.b.c.d, w.x.y.z
TAXII_CACHE_PORT int 6379 Port on which cache daemon/service is listening
TAXII_CACHE_TTL int 600 Amount of time a cache entry can idle in cache before removal/expiration. Measured in seconds.
TAXII_CACHE_MAX_ITEM_SIZE int 50m The maximum size (in bytes) per item that can be cached. Analogous to the memcached -I parameter which allows you to specify the maximum item size at runtime. It supports a unit postfix to allow for natural expression of item size.
TAXII_CACHE_NET_TIMEOUT int 6000 Specifies how long the TAXII server will wait for responses from Memcached. Measured in ms.
TAXII_CACHE_RECONNECT bool true Specifies whether the server should continue re-attempting to connect the cache host in the event there is a disconnect
TAXII_CACHE_MEM_SIZE int 4096 Sets the amount of memory allocated to memcached for object storage. ONLY USED BY MEMCACHED.
TAXII_CORS_ENABLED bool false Specifies whether CORS should be enabled on the server
TAXII_STIX_SRC_URL str http://localhost:3000 Specifies the address and port on which the Workbench REST API is listening.
TAXII_STIX_DATA_SRC str workbench Specifies how the server will source/ingest STIX data. At the moment, only 'workbench' is supported.
TAXII_WORKBENCH_AUTH_HEADER str dGF4aWktc2VydmVyOnNlY3JldC1zcXVpcnJlbA== Specifies the base64-encoded portion of the Authorization header that should be used on HTTP requests to the Workbench REST API.
TAXII_MONGO_URI str mongodb://localhost/taxii Specifies the URI of the MongoDB instance.
TAXII_LOG_LEVEL str info Default winston logging level. Conforms to RFC5424
TAXII_LOG_TO_FILE bool false Specifies whether the server should write logs to file (in addition to stdout)
TAXII_HTTPS_ENABLED bool true Specifies whether the server should use HTTPS (SSL/TLS)
TAXII_HYDRATE_ON_BOOT bool false Specifies whether the TAXII collector should begin database hydration at runtime.
TAXII_LOG_TO_HTTP_HOST str - Specifies the address (IP or FQDN) of a log listener. The server will attempt to send logs to this address if a value is set.
TAXII_LOG_TO_HTTP_PORT int - Specifies the port of the log listener
TAXII_LOG_TO_HTTP_PATH str - Specifies the HTTP endpoint of the log listener
TAXII_LOG_TO_SLACK_URL str - Specifies a Slack URL. The server will attempt to send logs to this address if a value is set.
TAXII_LOG_TO_SENTRY_DSN str - Specifies a Sentry Data Source Name (DSN). The server will attempt to send logs to this address if a value is set.
TAXII_SSL_PRIVATE_KEY str - Base64 encoded string containing the SSL/TLS private key.
TAXII_SSL_PUBLIC_KEY str - Base64 encoded string containing the SSL/TLS public key.

Authorization

If the TAXII server is configured to source STIX from the WorkbenchRepository, then the TAXII server must authenticate to the Workbench REST API using Basic Auth. There are two steps to enabling the TAXII server to authenticate to the Workbench REST API:

  1. Basic Auth must be enabled on the Workbench REST API and a service account & API key must be provisioned for the TAXII server. An example is provided here.

  2. On the TAXII server, TAXII_WORKBENCH_AUTH_HEADER must be set to the aforementioned encoded service account name and API key. The values must be base64 encoded from the string format service-name:api-key.

e.g., The Workbench REST API has configured a basic authorization service account for the TAXII server. The username is taxii-server and the key is secret-squirrel.

// rest-api-service-config.json

{
  "serviceAuthn": {
    "basicApikey": {
      "enable": true,
      "serviceAccounts": [
        {
          "name": "taxii-server",
          "apikey": "secret-squirrel",
          "serviceRole": "read-only"
        }
      ]
    }
  }
}

The Workbench REST API expects clients to authenticate with Authorization header Basic dGF4aWktc2VydmVyOnNlY3JldC1zcXVpcnJlbA==.

The string is a base-64 encoded string generated from the UTF-8 string value, taxii-server:secret-squirrel.

{
  "decoded": "taxii-server:secret-squirrel",
  "encoded": "dGF4aWktc2VydmVyOnNlY3JldC1zcXVpcnJlbA=="
}

The TAXII server is configured to send HTTP requests to the Workbench REST API with the following header:

{
  "Authorization": "Basic dGF4aWktc2VydmVyOnNlY3JldC1zcXVpcnJlbA=="
}

The base64 encoded portion of the authorization string is injected to the TAXII server using environment variable TAXII_WORKBENCH_AUTH_HEADER.

{
  Authorization: `Basic ${env.TAXII_WORKBENCH_AUTH_HEADER}`
}

HTTPS (TLS)

SSL/TLS can be enabled or disabled by setting the TAXII_HTTPS_ENABLED environment variable.

$ export TAXII_HTTPS_ENABLED=true   # enable HTTPS
$ export TAXII_HTTPS_ENABLED=false  # disable HTTPS

If enabled, a private key and a public key must be provided. The public and private keys can be provided in two ways:

  1. Set environment variables TAXII_SSL_PRIVATE_KEY and TAXII_SSL_PUBLIC_KEY. Both keys must be base64 encoded. A simple JS utility, encodePem.js, is provided in the project root directory to help convert your PEM files to base64 encoded strings. See the encodePem.js file for instructions on how to use it.
  2. Place the keys in {project-root}/config/. Name the file containing the public key public-certificate.pem and name the file containing the private key private-key.pem.

The server will prioritize option 1 over option 2 if both are set. In other words, if config/private-key.pem & config/public-certificate.pem exists, and TAXII_SSL_PRIVATE_KEY & TAXII_SSL_PRIVATE_KEY are set, then the former (config/*.pem) will take precedence.

Database

An instance of MongoDB is required in order to run the TAXII 2.1 server. By default, the server will query for STIX objects from the taxii database. The TAXII server will generate (and serve) TAXII resources based on the contents of the following two Mongo Collections:

  • attackObjects : Stores STIX resources
  • taxiicollections : Stores TAXII Collection resources

The taxii database is populated by the TAXII Collector, which can be initialized & run via npm run hydrate. Synchronization is handled by the HydrateModule and occurs every hour at the HH:00 and HH:30 minute mark (e.g., 12:00, 12:30, 1:00, 1:30).

The TAXII Collector can be configured to start hydrating the database as soon as it initializes by setting the TAXII_HYDRATE_ON_BOOT environment variable to true. This can be helpful in instances where the database is empty (and thus requires hydration) but the time is something like 12:45. In this example, rather than wait 15 minutes for the automated task scheduler to trigger, TAXII_HYDRATE_ON_BOOT can be used to force hydration to occur as soon as the TAXII collector initializes. Alternatively, if the database is already hydrated (and thus hydration is not needed at runtime), TAXII_HYDRATE_ON_BOOT can be disabled. This will restrict the task scheduler to only run at the HH:00 and HH:30 minute mark.

Clustering

The production instance of the TAXII server (npm run start:prod) is encapsulated in a PM2 cluster. Two independent processes are contained in the cluster:

  1. TAXII 2.1 Server (taxii21-server) : Initialized by src/main.ts, this process runs the user-facing web API and serves STIX data from an instance of MongoDB (as defined by the TAXII_MONGO_URI environment variable).

  2. TAXII Collector (taxii21-collector) : Initialized by src/hydrate.ts, this process is responsible for synchronizing and hydrating the Mongo database. This process is expected to run in the background and operates on a timer via the Nest.js native task scheduler. It retrieves STIX objects from an instance of Workbench (as determined by the TAXII_STIX_SRC_URL environment variable) and writes them to Mongo Collections that the taxii21-server process queries resources from.

/app # pm2 list
┌─────┬──────────────────────┬─────────────┬─────────┬─────────┬──────────┬────────┬──────┬───────────┬──────────┬──────────┬──────────┬──────────┐
│ id  │ name                 │ namespace   │ version │ mode    │ pid      │ uptime │ ↺    │ status    │ cpu      │ mem      │ user     │ watching │
├─────┼──────────────────────┼─────────────┼─────────┼─────────┼──────────┼────────┼──────┼───────────┼──────────┼──────────┼──────────┼──────────┤
│ 1   │ taxii21-collector    │ default     │ 0.9.0   │ cluster │ 19       │ 4h     │ 0    │ online    │ 0%       │ 96.0mb   │ root     │ disabled │
│ 0   │ taxii21-server       │ default     │ 0.9.0   │ cluster │ 18       │ 4h     │ 0    │ online    │ 0%       │ 129.4mb  │ root     │ disabled │
│ 2   │ taxii21-server       │ default     │ 0.9.0   │ cluster │ 32       │ 4h     │ 0    │ online    │ 0%       │ 129.5mb  │ root     │ disabled │
│ 3   │ taxii21-server       │ default     │ 0.9.0   │ cluster │ 39       │ 4h     │ 0    │ online    │ 0%       │ 129.2mb  │ root     │ disabled │
│ 4   │ taxii21-server       │ default     │ 0.9.0   │ cluster │ 46       │ 4h     │ 0    │ online    │ 0%       │ 129.0mb  │ root     │ disabled │
└─────┴──────────────────────┴─────────────┴─────────┴─────────┴──────────┴────────┴──────┴───────────┴──────────┴──────────┴──────────┴──────────┘

Logging

The TAXII server ships with a custom logger implementation that extends an instance of winston to the default Nest.js module. The logger is a Nest.js provider that uses the REQUEST injection scope in conjunction with the request-context module to correlate all log output to the user request. All inbound user HTTP request objects are assigned a unique ID.

In the following example, the server received an HTTP request on the discovery endpoint, GET /taxii2/. The first line indicates that the SetRequestIdMiddleware provider received the request and assigned random uuid a48bab11-e950-5e4f-9b9d-dcd9372ac3dd to it. The second and third lines show the request object flowing through the request pipeline. The same UUID appears in all log output, so we can easily trace how & where the user request flowed through the system.

[Nest] 1  - 06/08/2022, 2:15:28 PM     LOG [SetRequestIdMiddleware] New request: [a48bab11-e950-5e4f-9b9d-dcd9372ac3dd] GET /taxii2/ - application/taxii+json;version=2.1  insomnia/2022.3.0 ::ffff:172.17.0.1
[Nest] 1  - 06/08/2022, 2:15:28 PM   DEBUG [RootController] [a48bab11-e950-5e4f-9b9d-dcd9372ac3dd]  Received a discovery request
[Nest] 1  - 06/08/2022, 2:15:28 PM     LOG [ResLoggerMiddleware] Outgoing response: [a48bab11-e950-5e4f-9b9d-dcd9372ac3dd] 200 GET /taxii2/ 214 - insomnia/2022.3.0 ::ffff:172.17.0.1

Log Level

Environment variable TAXII_LOG_LEVEL is used to configure the embedded winston instance, which conveniently conforms to the severity ordering specified by RFC5424.

Logging levels in winston conform to the severity ordering specified by RFC5424: severity of all levels is assumed to be numerically ascending from most important to least important.

Each level is given a specific integer priority. The higher the priority the more important the message is considered to be, and the lower the corresponding integer priority. For example, as specified exactly in RFC5424 the syslog levels are prioritized from 0 to 7 (highest to lowest).

Logs sent to stdout cannot be disabled or filtered at this time.

The following values are supported by TAXII_LOG_LEVEL:

  • emerg
  • alert
  • crit
  • error
  • warning
  • notice
  • info (default)
  • debug

Log To File

Set environment variable TAXII_LOG_TO_FILE to true to configure the server to write logs to a file in addition to stdout. If TAXII_LOG_TO_FILE is not set, then the logger will not attempt to write logs to file.

When enabled, logs are written to a file named taxii_server_${timestamp}.log, e.g., taxii_server_2022-02-17.log.

Logs written to file by winston are formatted in JSON.

Log To HTTP

The embedded winston implementation is capable of sending logs to an HTTP endpoint. The logger can be configured to send logs to an HTTP listener by enabling the following three environment variables:

  • TAXII_LOG_TO_HTTP_HOST: the log listener's IP or FQDN
  • TAXII_LOG_TO_HTTP_PORT: the log listener's port
  • TAXII_LOG_TO_HTTP_PATH: the log listener's endpoint (e.g., /api/logs)

The TAXII server will attempt to enable the HTTP log forwarder when TAXII_LOG_TO_HTTP_HOST is set to a defined value.

Please note that logging to HTTP has not been fully tested. This feature is provided as is.

Log To Slack

The embedded winston implementation takes advantage of winston-slack-webhook-transport to forward logs to a Slack webhook. The full webhook URL can be configured via environment variable TAXII_LOG_TO_SLACK_URL. If TAXII_LOG_TO_SLACK_URL is not set, then the logger will not attempt to send logs to Slack.

Please note that logging to Slack has not been fully tested. This feature is provided as is.

Log To Sentry

The embedded winston implementation takes advantage of winston-transport-sentry-node to forward logs to Sentry. The Sentry DSN path (e.g., 'https://******@sentry.io/12345') can be configured via environment variable TAXII_LOG_TO_SENTRY_DSN. If TAXII_LOG_TO_SENTRY_DSN is not set, then the logger will not attempt to send logs to Sentry.

Please note that logging to Sentry has not been fully tested. This feature is provided as is.

Caching

The TAXII server provides a dynamic cache provider that supports two types of caches:

  1. An in-memory cache (default)
  2. Memcached

Caching is used by the WorkbenchModule to store responses from the Workbench REST API.

Memcached

The TAXII server can be configured to use Memcached by setting environment variable TAXII_CACHE_TYPE to memcached.

$ export TAXII_CACHE_TYPE=memcached

While external cache implementations are outside the scope of the TAXII server, a containerized instance of memcached can easily be spun-up by using the Docker Compose template located in the attack-workbench-deployment repository.

The following configuration parameters can be set via environment variables when memcached is enabled:

  • TAXII_CACHE_HOST: The IP address or FQDN of the Memcached server
  • TAXII_CACHE_PORT: The port on which the Memcached server is listening
  • TAXII_CACHE_TTL: The duration of time that an item should remain in the cache before removal
  • TAXII_CACHE_MAX_ITEM_SIZE: The maximum size of each item allowed in the cache. This value also sets the corresponding server-side setting (i.e., Maps to the memcached -I flag.)
  • TAXII_CACHE_MEM_SIZE: Sets the amount of memory allocated to memcached for object storage. (e.g., Maps to the memcached -m flag.) This setting only applies to the Memcached server; not the TAXII app!

Note that the TAXII application can only map to one instance of Memcached. Multiple Memcached servers cannot be used.

The following configuration parameters can be set via environment variables when the in-memory (default) cache is enabled:

  • TAXII_CACHE_TTL: The duration of time that an item should remain in the cache before removal

Scripts

package.json contains a number of scripts that can be used to perform recurring tasks.

  • prebuild: deletes the entire dist/ directory if it exists.
  • build: transpiles the TypeScript code to JavaScript in the dist/ directory.
  • start: starts the server in a single Node.js process.
  • hydrate: starts the TAXII collector process.
  • start:prod: starts the server in a PM2 cluster. References ecosystem.config.js for instructions.
  • start:dev: starts the server in "hot-reload" mode. Useful for development.
  • test:e2e: starts the end-to-end HTTP tests to ensure that all available TAXII endpoints are working as expected. Note that the E2E test suite requires a live, pre-populated database to run correctly.
  • test: executes a series of unit tests to ensure that all Nest.js components/modules are working as expected.

A few additional scripts are provided to simplify various aspects of the deployment process:

  • docker_build.sh is a simple wrapper script that makes it easier to [re]build the Docker image.

  • run.sh cleans, builds, and runs a containerized instance of the TAXII server. Details are described in the Build from source section of the README file.

  • encode.js is described in the HTTPS section of this document.