NGINX App Protect DoS Directives and Policy

Introduction

NGINX directives are written in the nginx.conf file and are used to configure specific modules of NGINX.
NGINX App Protect DoS has its own directives (which follows the same rules of all other directives) and are used to enable and configure it.

The table below summarizes all the NGINX App Protect DoS directives.

Although only the first directive is mandatory (used for enabling NGINX App Protect DoS), it is advised to use as many directives as possible in order to make use of the product’s monitoring and application health detection capabilities.
When reloading NGINX after inserting the directives, pay attention to any errors or warnings that might be written to the NGINX error log.

Directives table

This is a summary of all NGINX App Protect DoS directives. You can find the description of each below.

Directives Info

Enable directive (app_protect_dos_enable)

Enables/disables App Protect DoS module in the relevant block/s.
It can be written in the following contexts: location/server/http.

The derived blocks/contexts also inherit the directive. For example: A directive written in http context will be considered as if written also in all of the http’s server blocks and their location blocks.

In case of multiple directives in different contexts, the derived overwrites the base’s directive.

Example:

app_protect_dos_enable on;

Policy directive (app_protect_dos_policy_file)

This is the path to the JSON policy file which includes the product’s configuration parameters. It can be written in the following contexts: location/server/http.

The directive is optional. If not inserted then default path will be used, which is /etc/app_protect_dos/BADOSDefaultPolicy.json.

If the configuration file doesn’t exist or its attributes are invalid, default values will be used.

BADOSDefaultPolicy.json:

{
    "mitigation_mode": "standard",
    "signatures": "on",
    "bad_actors": "on",
    "automation_tools_detection": "on",
    "tls_fingerprint" : "on"
}

Example:

app_protect_dos_policy_file /etc/app_protect_dos/BADOSPolicy.json;

Service Name directive (app_protect_dos_name)

This is the VS (protected object) name, it should be unique and is used to identify the VS in the logs. It can be used in location/server/http blocks.

Directive is optional. If not written, then each protected object (VS) will have an auto-generated name according to the following syntax:

line_number-server_name:seq-location_name

For example: 30-backend:1-/abc

• line number: the line number of the server block (server {) in the nginx.conf file (i.e. 30)
  
• server name: taken from directive server_name (i.e. backend)
  seq: 0 for server block, increments for each location block. i.e. VS created from server block will have 0 and VS’s from location blocks will be 1,2,3,… (i.e. 1)
• location name: the name of the location (i.e. /abc)

Example:

app_protect_dos_name po-example;

Monitor directive (app_protect_dos_monitor)

This is how NGINX App Protect DoS monitors the stress level of the protected object. The monitor requests are sent from localhost (127.0.0.1).

It can be used in location/server/http blocks.

For NGINX App Protect DoS v2+

Monitor directive has three named arguments - uri, protocol and timeout. The first is mandatory, the rest are optional.

• URI - is the destination to the desired protected object in the nginx.conf: scheme://server_name:port/location.

  Note:
  For gRPC, a certain method should be specified in the location. For example, /RouteGuide/GetFeature.

• Protocol - determines if the server listens on http1 / http2 / grpc. The default is http1.

• Timeout - determines how long (in seconds) should NGINX App Protect DoS wait for a response. Default is 10 seconds for http1/http2 and 5 seconds for grpc.

Monitor directive is mandatory for http2/grpc and optional for http1.

For NGINX App Protect DoS v1:

Monitor directive has one argument which is the uri to be monitored.

The argument is a destination to the desired protected object in the nginx.conf: scheme://server_name:port/location.

Examples:

1. http1 on port 80:

listen 80;
server_name serv;
 
location / {
    # Protected Object is defined here
}

app_protect_dos_monitor uri=http://serv:80/;

For NGINX App Protect DoS v1: app_protect_dos_monitor http://serv:80/;

2. http2 over SSL

listen 443 http2 reuseport ssl;
server_name serv;
 
location / {
    # Protected Object is defined here
}

app_protect_dos_monitor uri=https://serv:443/ protocol=http2 timeout=5;

3. gRPC (over http2) on port 50051

listen 50051 http2 reuseport;
server_name my_grpc;
 
location /routeguide. {
    # Protected Object is defined here
}

app_protect_dos_monitor uri=http://my_grpc:50051/routeguide.RouteGuide/GetFeature protocol=grpc;

Security log enable directive (app_protect_dos_security_log_enable)

Enable/Disable App Protect DoS security logger. It can be used in location/server/http blocks.

Directive is optional. If not written, then logger is disabled.

Example:

app_protect_dos_security_log_enable on;

Security log directive (app_protect_dos_security_log)

This directive has two string arguments.

First argument is the configuration file path, i.e. /etc/app_protect_dos/log-default.json.

Second argument is the destination (the location which the events will be sent to). The destination can be one of three options:

• syslog:server={ip}:{port}, i.e. syslog:server=1.2.3.4:3000
• stderr (default)
• {absolute_file_path}, i.e. /shared/dos_sec_logger.log

Implemented according to: NGINX App Protect DoS Security Log

Note:

• When using stderr, make sure that the process admd is not redirecting the stderr output to file.
• When using the Docker entrypoint.sh startup script from the admin guide, make sure that it doesn’t redirect stderr.

Examples:

• Syslog:

app_protect_dos_security_log "/etc/app_protect_dos/log-default.json" syslog:server=1.2.3.4:5000;

• File:

app_protect_dos_security_log "/etc/app_protect_dos/log-default.json" /shared/logger.log;

• Stderr:

app_protect_dos_security_log "/etc/app_protect_dos/log-default.json" stderr;

While /etc/app_protect_dos/log-default.json is:

{
    "filter": {
        "traffic-mitigation-stats": "all",
        "bad-actors": "top 10",
        "attack-signatures": "top 10"
    },
    "content": {
        "format": "splunk"
    }
}

Liveness probe directive (app_protect_dos_liveness)

This directive has 3 arguments.

Note:
Second and Third arguments are optional; if one or more is not written, the default will take place.

If liveness is enabled, a request with URI and PORT that matches the probe configuration (i.e. /app_protect_dos_liveness:8090) will be answered with RC 200 “Alive” by our NGINX module, without being counted or pass to other handlers nor the backend server.

Any other response will indicate that our NGINX module (NGINX App Protect DoS) has not received the request (possibly means that NGINX is down).

Example:

app_protect_dos_liveness on uri:/liveness port:8090;

Readiness probe directive (app_protect_dos_readiness)

This directive has 3 arguments.

Note:
Second and Third arguments are optional; if one or more is not written, the default will take place.

If readiness is enabled, a request with URI and PORT that matches the probe configuration (i.e. /app_protect_dos_readiness:8090) will be answered with RC 200 “Ready” or RC 503 “Not Ready” by our NGINX module, without being counted or pass to other handlers nor the backend server.

Any other response will indicate that our NGINX module (NGINX App Protect DoS) has not received the request (possibly means that NGINX is down).

RC 200 “Ready” will occur if two conditions are met:

1. NGINX worker successfully connected to the global shared memory block
2. ADMD process is running (and not stuck)

Example:

app_protect_dos_readiness on uri:/readiness port:8090;