add the feature of tcp proxy with nginx, with health check and status monitor
add the feature of tcp proxy with nginx, with health check and status monitor
nginx_tcp_proxy_module - support TCP proxy with Nginx
Download the latest stable version of the release tarball of this module from github (http://github.com/yaoweibin/nginx_tcp_proxy_module)
Grab the nginx source code from nginx.org (http://nginx.org/), for example, the version 1.20.2 (see nginx compatibility), and then build the source with this module:
$ wget 'http://nginx.org/download/nginx-1.20.2.tar.gz'
$ tar -xzvf nginx-1.20.2.tar.gz
$ cd nginx-1.20.2/
$ patch -p1 < /path/to/nginx_tcp_proxy_module/tcp.patch
$ ./configure --add-module=/path/to/nginx_tcp_proxy_module
$ make
$ make install…This module actually include many modules:
ngx_tcp_module,
ngx_tcp_core_module,
ngx_tcp_upstream_module,
ngx_tcp_proxy_module,
ngx_tcp_websocket_module,
ngx_tcp_ssl_module,
ngx_tcp_upstream_ip_hash_moduleAll these modules work together to support TCP proxy with Nginx. I also added other features:
ip_hash,
upstream server health check,
status monitorThe motivation of writing these modules is Nginx's high performance and robustness. At first, I developed this module just for general TCP proxy. And now, this module is frequently used in websocket reverse proxying.
Note, You can't use the same listening port with HTTP modules.
syntax: tcp {...}
default: none
context: main
description: All the tcp related directives are contained in the tcp
block.
syntax: server {...}
default: none
context: tcp
description: All the specific server directives are contained in the
server block.
syntax: listen address:port [ bind | ssl | default]
default: none
context: server
description: The same as listen (http://wiki.nginx.org/NginxMailCoreModule#listen). The parameter of
default means the default server if you have several server blocks with
the same port.
syntax: access_log path [buffer=size] | off
default: access_log logs/tcp_access.log
context: tcp, server
description: Set the access.log. Each record's format is like this:
log_time worker_process_pid client_ip host_ip accept_time upstream_ip
bytes_read bytes_write
2011/08/02 06:19:07 [5972] 127.0.0.1 0.0.0.0:1982 2011/08/02 06:18:19
172.19.0.129:80 80 236305log_time - The current time when writing this log. The log action is called when the proxy session is closed. worker_process_pid - the pid of worker process client_ip - the client ip host_ip - the server ip and port accept_time - the time when the server accepts client's connection upstream_ip - the upstream server's ip bytes_read - the bytes read from client bytes_write - the bytes written to clientsyntax: allow [ address | CIDR | all ]
default: none
context: server
description: Directive grants access for the network or addresses indicated.
syntax: deny [ address | CIDR | all ]
default: none
context: server
description: Directive grants access for the network or addresses indicated.
syntax: so_keepalive on|off
default: off
context: main, server
description: The same as so_keepalive
(http://wiki.nginx.org/NginxMailCoreModule#so_keepalive).
syntax: tcp_nodelay on|off
default: on
context: main, server
description: The same as tcp_nodelay
(http://wiki.nginx.org/NginxHttpCoreModule#tcp_nodelay).
syntax: timeout milliseconds
default: 60000
context: main, server
description: set the timeout value with clients.
syntax: server_name name
default: The name of the host, obtained through gethostname()`
context: tcp, server
description: The same as server_name
(http://wiki.nginx.org/NginxMailCoreModule#server_name). You can
specify several server name in different server block with the same
port. They can be used in websocket module.
syntax: resolver address
default: none
context: tcp, server
description: DNS server
syntax: resolver_timeout time
default: 30s
context: tcp, server
description: Resolver timeout in seconds.
syntax: upstream {...}
default: none
context: tcp
description: All the upstream directives are contained in this block. The upstream server will be dispatched with round robin by default.
syntax: server name [parameters]
default: none
context: upstream
description: Most of the parameters are the same as server
(http://wiki.nginx.org/NginxHttpUpstreamModule#server). Default port is 80.
syntax:
check interval=milliseconds [fall=count][rise=count]
[timeout=milliseconds] [type=tcp|ssl_hello|smtp|mysql|pop3|imap]default: none, if parameters omitted, default parameters are
interval=30000 fall=5 rise=2 timeout=1000
context: upstream
description: Add the health check for the upstream servers. At present,the check method is a simple tcp connect.
The parameters' meanings are:
interval: the check request's interval time. fall(fall_count): After fall_count check failures, the server is marked down. rise(rise_count): After rise_count check success, the server is marked up. timeout: the check request's timeout. type: the check protocol type:tcp is a simple tcp socket connect and peek one byte.ssl_hello sends a client ssl hello packet and receives the server ssl hello packet.http sends a http request packet, receives and parses the http response to diagnose if the upstream server is alive.smtp sends a smtp request packet, receives and parses the smtp response to diagnose if the upstream server is alive. The response begins with '2' should be an OK response.mysql connects to the mysql server, receives the greeting response to diagnose if the upstream server is alive.pop3 receives and parses the pop3 response to diagnose if the upstream server is alive. The response begins with '+' should bean OK response.imap connects to the imap server, receives the greeting response to diagnose if the upstream server is alive.syntax: check_http_send http_packet
default: "GET / HTTP/1.0\r\n\r\n"
context: upstream
description: If you set the check type is http, then the check function will sends this http packet to check the upstream server.
syntax: check_http_expect_alive [ http_2xx | http_3xx | http_4xx | http_5xx ]
default: http_2xx | http_3xx
context: upstream
description: These status codes indicate the upstream server's http response is OK, the backend is alive.
syntax: check_smtp_send smtp_packet
default: "HELO smtp.localdomain\r\n"
context: upstream
description: If you set the check type is smtp, then the check function will sends this smtp packet to check the upstream server.
syntax: check_smtp_expect_alive [smtp_2xx | smtp_3xx | smtp_4xx | smtp_5xx]
default: smtp_2xx
context: upstream
description: These status codes indicate the upstream server's smtp response is OK, the backend is alive.
syntax: check_shm_size size
default: (number_of_checked_upstream_blocks + 1) * pagesize
context: tcp
description: If you store hundreds of servers in one upstream block, the shared memory for health check may be not enough, you can enlarged it by this directive.
syntax: tcp_check_status
default: none
context: location
description: Display the health checking servers' status by HTTP. This directive is set in the http block.
The table field meanings are:
Index: The server index in the check table Name : The upstream server name Status: The marked status of the server. Busyness: The number of connections which are connecting to the server. Rise counts: Count the successful checking Fall counts: Count the unsuccessful checking Access counts: Count the times accessing to this server Check type: The type of the check packetsyntax: busyness
default: none
context: upstream
description: the upstream server will be dispatched by backend servers busyness.
syntax: ip_hash
default: none
context: upstream
description: the upstream server will be dispatched by ip_hash.
syntax: proxy_pass host:port
default: none
context: server
description: proxy the request to the backend server. Default port is 80.
syntax: proxy_buffer size
default: 4k
context: tcp, server
description: set the size of proxy buffer.
syntax: proxy_connect_timeout miliseconds
default: 60000
context: tcp, server
description: set the timeout value of connection to backends.
syntax: proxy_read_timeout miliseconds
default: 60000
context: tcp, server
description: set the timeout value of reading from backends.
syntax: proxy_send_timeout miliseconds
default: 60000
context: tcp, server
description: set the timeout value of sending to backends.
syntax: websocket_pass [path] host:port
default: none
context: server
description: proxy the websocket request to the backend server. Default port is 80. You can specify several different paths in the same server block.
syntax: websocket_buffer size
default: 4k
context: tcp, server
description: set the size of proxy buffer.
syntax: websocket_connect_timeout miliseconds
default: 60000
context: tcp, server
description: set the timeout value of connection to backends.
syntax: websocket_read_timeout miliseconds
default: 60000
context: tcp, server
description: set the timeout value of reading from backends. Your timeout will be the minimum of this and the timeout parameter, so if you want a long timeout for your websockets, make sure to set both parameters.
syntax: websocket_send_timeout miliseconds
default: 60000
context: tcp, server
description: set the timeout value of sending to backends.
The default config file includes this ngx_tcp_ssl_module. If you want to just compile nginx without ngx_tcp_ssl_module, copy the ngx_tcp_proxy_module/config_without_ssl to ngx_tcp_proxy_module/config, reconfigrure and compile nginx.
syntax: ssl [on|off]
default: ssl off
context: tcp, server
Enables SSL for a server.
syntax: ssl_certificate file
default: ssl_certificate cert.pem
context: tcp, server
This directive specifies the file containing the certificate, in PEM format. This file can contain also other certificates and the server private key.
syntax: ssl_certificate_key file
default: ssl_certificate_key cert.pem
context: tcp, server
This directive specifies the file containing the private key, in PEM format.
syntax: ssl_client_certificate file
default: none
context: tcp, server
This directive specifies the file containing the CA (root) certificate, in PEM format, that is used for validating client certificates.
syntax: ssl_dhparam file
default: none
context: tcp, server
This directive specifies a file containing Diffie-Hellman key agreement p
No open issues yet, or sync has not completed.