Enable to check for illegal characters in the Host: line of the HTTP header, such as null characters or encoded characters.

For example, 0x0 or %00* are illegal.

Attack log messages contain Illegal Host Name when this feature detects an invalid host name.

Enable to check for invalid HTTP version numbers. Currently, the only valid version strings are HTTP/0.9, HTTP/1.0 or HTTP/1.1.

Attack log messages contain Illegal HTTP Version when this feature detects an invalid HTTP version number.

Specifies the maximum acceptable size in bytes of the HTTP body.

For requests that use the HTTP POST method, this typically includes parameters submitted by HTML form inputs. In the case of file uploads, this can normally be many megabytes. For most simple forms, however, the body should be only a few kilobytes in size at maximum.

Attack log messages contain Body Length Exceeded when this feature detects a body size buffer overflow attempt.

Specifies the maximum acceptable number of cookies in an HTTP request.

Attack log messages contain Too Many Cookies in Request when this feature detects a cookie count buffer overflow attempt.

Specifies the maximum acceptable number of Range: lines in each HTTP header. The default value is 5.

Attack log messages contain Too Many Range Headers when this feature detects too many Range: header lines.

Tip: Some versions of Apache are vulnerable to a denial of service (DoS) attack on this header, where a malicious client floods the server with many Range: headers. The default value is appropriate for un-patched versions of Apache 2.0 and Apache 2.1.

Enable to inspect the request for:

• syntax errors
• exceeding the maximum buffer size allowed by FortiWeb’s HTTP parser

Errors and buffer overflows can cause problems in web servers that do not handle them gracefully. Such problems can lead to security vulnerabilities.

Attack log messages contain Too Many Parameters or Too Many Flash Parameters or another message that indicates the specific cause when this feature detects a request with parser errors or a FortiWeb buffer overflow attempt.

Caution: Fortinet strongly recommends to enable this option unless large requests/parameters are required by the web application. If part of a request is too large for its scan buffer, FortiWeb cannot scan it for attacks. It also cannot perform rewrites. Unless you configure it to block, FortiWeb allows oversized requests to pass through without scanning or rewriting. This could allow padded attacks to pass through, and rewriting to be skipped.

If feasible, instead of disabling this option:

• Enlarge the scan buffer for each parameter (see http-cachesize in the FortiWeb CLI Reference). Requests larger than the buffer will be flagged as potentially malformed by FortiWeb’s parser, causing FortiWeb to block normal requests (i.e. false positives). For more buffer specifications, see Buffer hardening.
• Disable this setting only for URLs that require oversized parameters (see Configuring HTTP protocol constraint exceptions)

Enable to detect traffic that uses the WebSocket TCP-based protocol.

Because FortiWeb acts as a pure socket proxy for WebSocket traffic, it cannot apply security features to it.

Specifies the maximum acceptable size in bytes of all HTTP header lines.

Attack log messages contain Total Size of All Headers Too Large when this feature detects a header size buffer overflow attempt.

Enable to check for invalid HTTP request methods according to RFC 2616 or RFC 4918. Any method not defined in these RFCs — including misspellings like GETT as well as other HTTP extension methods (e.g. CalDAV) like MKCALENDAR — are considered invalid.

Attack log messages contain Illegal HTTP Method when this feature detects an invalid HTTP request method.

Specifies the maximum acceptable length in bytes of the entire HTTP request, including both headers and body.

Attack log messages contain HTTP Request Length Exceeded when this feature detects an excessively large HTTP request.

Specifies the maximum acceptable number of lines in the HTTP header.

Attack log messages contain Too Many Headers when this feature detects a header line count buffer overflow attempt.

Specifies the total maximum acceptable length in bytes of all parameters, including their names and values, in the URL. Parameters usually appear after a ?, such as:

/url?parameter1=value1&parameter2=value2

The count does not include:

• Question mark ( ? ), ampersand ( & ), and equal ( = ) characters are not included.
• Parameters in the HTTP body, which can occur with HTTP POST requests. For these parameters, configure Total Body Parameters Length or Body Length instead.

Attack log messages contain Total URL Parameters Length Exceeded when this feature detects a URL parameter line length buffer overflow attempt.

Specifies the total maximum acceptable size in bytes of all the parameters in the HTTP body of HTTP POST requests.

Question mark ( ? ), ampersand ( & ), and equal ( = ) characters are not included.

Attack log messages contain Total Body Parameters Length Exceeded when this feature detects a total parameter size buffer overflow attempt.

Specifies the maximum number of parameters in the URL. The maximum number is 1024.

It does not include parameters in the HTTP body, which can occur with HTTP POST requests.

Attack log messages contain Too Many Parameters in Request when this feature detects a URL parameter count buffer overflow attempt.

The default is 128.

Specifies the maximum acceptable length in bytes of the request body. Length is determined by comparing this limit with the value of the Content-Length: field in the HTTP header.

Attack log messages contain Content Length Exceeded when this feature detects a content length buffer overflow attempt.

Tip: RPC requests’ content length often do not match their own Content-Length: header. Attackers may also intentionally craft mismatching Content-Length: headers in an attempt to cloak buffer overflows. For those cases, use other limits instead or in addition, such as Body Length and Limiting file uploads.