w3wsrmn at 2009-6-20 8:09 wrote:
> A look at the script reveals it keeps connections open with invalid headers (note the appended "\r\n"):
>
> "GET /$rand HTTP/1.1\r\n"
> . "Host: $sendhost\r\n"
> . "User-Agent: Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 5.1; Trident/4.0; .NET CLR 1.1.4322; .NET CLR 2.0.503l3; .NET CLR 3.0.4506.2152; .NET CLR 3.5.30729; MSOffice 12)\r\n"
> . "Content-Length: 42\r\n";
>
> As by default the (undocumented?) ignore_invalid_headers directive is enabled in nginx, isn't this attack a non-issue, unless one disables the directive?
>
> Sending such headers to an nginx server with the directive enabled results in a "400 Bad Request".
>
>
When using telnet to send above header, I received the 400 response.
But when I tested the slowloris.pl script in nginx_0.7.59. The
ignore_invalid_headers directive is useless, Nginx treate the
header_line 'X-a: b\r\n' as valid header.
The debug log is like this:

2009/06/22 16:58:58 [debug] 25864#0: *1 accept: 172.19.1.209 fd:9
2009/06/22 16:58:58 [debug] 25864#0: *1 event timer add: 9: 60000:120682241
2009/06/22 16:58:58 [debug] 25864#0: *1 epoll add event: fd:9 op:1 ev:80000001
.
.
.
2009/06/22 16:58:58 [debug] 25864#0: *1 http process request line
2009/06/22 16:58:58 [debug] 25864#0: *1 recv: fd:9 236 of 1024
2009/06/22 16:58:58 [debug] 25864#0: *1 http request line: "GET / HTTP/1.1"
2009/06/22 16:58:58 [debug] 25864#0: *1 http uri: "/"
2009/06/22 16:58:58 [debug] 25864#0: *1 http args: ""
2009/06/22 16:58:58 [debug] 25864#0: *1 http exten: ""
2009/06/22 16:58:58 [debug] 25864#0: *1 http process request header line
2009/06/22 16:58:58 [debug] 25864#0: *1 http header: "Host: edu-9.space.163.org"
2009/06/22 16:58:58 [debug] 25864#0: *1 http header: "User-Agent: Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 5.1; Trident/4.0; .NET CLR 1.1.4322; .NET CLR 2.0.503l3; .NET CLR 3.0.4506.2152; .NET CLR 3.5.30729; MSOffice 12)"
2009/06/22 16:58:58 [debug] 25864#0: *1 http header: "Content-Length: 42"
2009/06/22 16:58:58 [debug] 25864#0: *1 recv: fd:9 -1 of 788
2009/06/22 16:58:58 [debug] 25864#0: *1 recv() not ready (11: Resource temporarily unavailable
.
.
.
2009/06/22 16:58:58 [debug] 25864#0: *1 http process request header line
2009/06/22 16:58:58 [debug] 25864#0: *1 recv: fd:9 8 of 788
2009/06/22 16:58:58 [debug] 25864#0: *1 http header: "X-a: b"
2009/06/22 16:58:58 [debug] 25864#0: *1 recv: fd:9 -1 of 780
2009/06/22 16:58:58 [debug] 25864#0: *1 recv() not ready (11: Resource temporarily unavailable)
.
.
.
2009/06/22 16:59:48 [debug] 25864#0: *1 http process request header line
2009/06/22 16:59:48 [debug] 25864#0: *1 recv: fd:9 8 of 780
2009/06/22 16:59:48 [debug] 25864#0: *1 http header: "X-a: b"
2009/06/22 16:59:48 [debug] 25864#0: *1 recv: fd:9 -1 of 772
2009/06/22 16:59:48 [debug] 25864#0: *1 recv() not ready (11: Resource temporarily unavailable)
.
.
.
2009/06/22 16:59:58 [debug] 25864#0: *1 event timer del: 9: 120682241
2009/06/22 16:59:58 [debug] 25864#0: *1 http process request header line
2009/06/22 16:59:58 [info] 25864#0: *1 client timed out (110: Connection timed out) while reading client request headers, client: 172.19.1.209, server: _, request: "GET / HTTP/1.1", host: "edu-9.space.163.org"
2009/06/22 16:59:58 [debug] 25864#0: *1 http close request
2009/06/22 16:59:58 [debug] 25864#0: *1 http log handler
2009/06/22 16:59:58 [debug] 25864#0: *1 close http connection: 9

The default timeout value is 60 seconds. But you can set with client_header_timeout(http://wiki.nginx.org/NginxHttpCoreModule#client_header_timeout), This directive is much useful。

I think Nginx is also effected by such DoS attack, but much stronger than apache.

--
Weibin Yao