|
|
| External Application |
LiteSpeed server can forward requests to external applications to
process and generate dynamic content.
Since 2.0, LiteSpeed server supports seven types of external
applications: CGI, FastCGI, Web Server, Servlet Engine, LSAPI
Application, Load Balancer and Piped Logger.
- CGI stands for Common Gateway Interface and the current standard is CGI/1.1.
CGI applications run in standalone processes. A CGI process is started
for each request and will exit after finishing the request.
- FastCGI is a fast, open, and secure web server interface that solves the
performance problems inherent from CGI without introducing the overhead and
complexity of proprietary APIs (Application Programming
Interfaces). For more information, please visit http://www.fastcgi.com
LiteSpeed web server defines two types of FastCGI applications:
Local FastCGI and remote FastCGI. There is no real difference
between them except how they are started. Local FastCGI refers to a
FastCGI application that is started automatically by the web server
on demand. Remote FastCGI refers to a FastCGI application that is
not started by the web server and it can run either on a remote
machine or on the same machine. In addition to generating dynamic
response (RESPONDER role), a FastCGI application can be used for
authentication (AUTHORIZER role) as well.
- Web Server is a server application that supports HTTP 1.0/1.1
protocols. LiteSpeed web server can serve as a transparent reverse proxy
to accelerate and protect any web server and application
server. When communicating with backend web servers, LiteSpeed will
always try to utilize "persistent connection" in order to achieve
highest performance.
- Servlet Engine is an application that implements Java Servlet and
Java Server Pages, like Apache Tomcat and Jetty. LiteSpeed can communicate
with Servlet engine through AJPv1.3 (Apache JServ Protocol version 1.3).
Servlet engine can only be started outside of LiteSpeed and can be setup to run on a
remote machine.
- LSAPI (LiteSpeed Server API) Application is an application that communicates with
LiteSpeed Web Server through LSAPI, which is a fast, scalable server interface
specfically designed for LiteSpeed Web Server to achieve maximum efficiency.
LSAPI has similar mechanism to FastCGI but LSAPI application can be up to 15%
faster than a comparable FastCGI application.
For the most demanding web applicatin, LSAPI is highly
recommended over FastCGI.
Our tests show that LSAPI PHP, using LiteSpeed SAPI, is
up to 30% faster than FastCGI PHP.
For more information about LSAPI and PHP Litespeed SAPI please visit
http://www.litespeedtech.com/lsapi/.
-
Load balancer is a virtual external application, which only assigns
requests to backend worker applications based on work load. Worker
applications can be CGI, FastCGI, web server, Servlet engine and LSAPI
applications. Different types of applications can
be mixed together under one load balancer. Load balancer and Piped
Loger cannot be used as worker applications.
- Piped Logger is an application that can process access
log in real-time, for examples, writing access log into database, performing
real-time statistics calculation, etc.
Unlike CGI application, FastCGI, LSAPI application and
Servlet Engine do not exit after processing a request: the process will stay/persist and serve
as many requests as it can. The socket connection can be reused for multiple requests as well.
There are two ways to use external applications in LiteSpeed web server:
- As a handler of a specific context;
- As a handler of a script language via file suffix binding.
FastCGI application, LSAPI application, web server, Servlet Engine
and Load Balancer have to be defined in External Application
section before being used in script handler or context configuration.
However, it is not necessary to define CGI application here.
External Applications can be configured at server level or at virtual host level.
When configured at server level, it can be used by all virtual hosts; when configured at
virtual host level, it can only be used by that virtual host.
|
| |
| Table of Contents |
|
| |
| Type |  |
|
Description: Specifies the type of external application. Choose from
- FastCGI: a FastCGI application with Responder
role.
- FastCGI Authorizer: a FastCGI application with
Authorizer role
- Servlet Engine: a Servlet engine with AJPv13 connector,
such as tomcat.
- Web Server: a web server or application server that
support HTTP protocol.
- LSAPI App: an application that communicate with web server
in LSAPI protocol.
- Load Balancer: a virtual application that can balance load
among worker applications.
- Piped Logger: an application that can process access log
entries received on its STDIN stream.
|
| Syntax: Select from drop down list |
|
| Name | |
|
| Description: A unique name for an external application.
You will refer it by this name when you use it in other parts of the configuration.
|
|
| Address | |
|
| Description: A unique socket address used by the external application.
IPv4 socket and Unix Domain Socket (UDS) are supported.
IPv4 socket can be used for communication over the network,
but UDS can only be used when the external application residing on the same machine.
|
Syntax: IPv4Addr:port or UDS://path
Example: 127.0.0.1:5434, UDS://tmp/lshttpd/php.sock. |
Tips: [Security] If the external application runs on the same machine,
UDS is preferred. If you have to use IPv4 socket,
set the IP part to localhost or 127.0.0.1,
thus the external application is inaccessible from other machines.
[Performance] Unix Domain Socket generally provides higher performance than an IPv4 socket. |
|
| Max Connections | |
|
| Description: Specifies the maximum concurrent connections that can be established
between the web server and an external application. This setting controls how
many requests can be processed concurrently by an external application.
However, the real limit also depends on the external application itself.
Set this value higher will not help if the external application is not fast enough
or cannot scale to large number of concurrent requests.
|
| Syntax: Integer number |
| Tips: [Performance] Setting a high value does not directly translate to higher performance.
Set the limit to a value that will not overload the the external
application will provide the best performance/throughput. |
|
| Environment | |
|
| Description: Specifies some extra environment variables for the external application.
|
| Syntax: key=value multiple variables can be separated by "ENTER" |
|
| Initial Request Timeout | |
|
| Description: Specifies the maximum time the web server waits for the external
application to respond to the first request over a new established connection.
If the web server does not receive any data from the external application
within this timeout limit, it will mark this connection as bad. This
helps to identify communication problem with external application as
quickly as possible. If some requests take longer time to
process, increase this limit to avoid "503" error messages.
|
| Syntax: Integer number |
|
| Retry Timeout | |
|
| Description: Specifies the period of time that the web server waits before
retrying the external application that had prior communication problem.
|
| Syntax: Integer number |
|
| Persistent Connection | |
|
| Description: Specifies whether to keep the connection open after a request
has been processed. Persistent connection can increase performance
but some FastCGI external applications does not support persistent connection
fully. In order to run Ruby On Rails via a FastCGI interface, persistent
connection needs to be turned off. Default is on. Currently, this
option is only effective for a FastCGI external application.
|
| Syntax: Select from radio box |
|
| Keep Alive Timeout | |
|
| Description: Specifies the maximum time to keep an idle persistent connection
open. When set to -1, the connection never timeout. When set to >=0,
the connection will be closed after the timeout has been reached.
|
| Syntax: int |
|
| Response Buffering | |
|
| Description: Specifies whether to buffer response received from external
applications. If Apache's Non-Parsed-Header mode is available enabled, buffering is turned off for response with full HTTP headers.
|
| Syntax: Select from drop down list |
|
| Auto Start | |
|
| Description: Specifies whether you want the web server to start the
application automatically. Only local FastCGI and LSAPI application
running on the same machine can be started automatically. The IP in the
Address must be a local IP.
|
| Syntax: Select from radio box |
|
| Max Idle Time | |
|
| Description: Specifies the maximum idle time before an external application
is stopped by the server. When set to -1, the external application
will not be stopped by the server. Default value is -1. This feature
allows resources used by idle applications to be freed. It is
especially useful in mass hosting environment when you need to define
many applications running in setuid mode for the sake of maximum
security.
|
| Syntax: Select from radio box |
|
| Command | |
|
| Description: Specifies the full command line including parameters to execute
the external application. Required value if
Auto Start is enabled. A parameter should be quoted with a
double or single quote if the parameter contains space or tab characters.
|
| Syntax: Full path to the executable with optional parameters. |
|
| Back Log | |
|
| Description: Specifies the backlog of the listening socket. Required if Auto Start is enabled.
|
| Syntax: Integer number |
|
| Instances | |
|
| Description: Specifies the maximum instances of the external application the server will
create. It is required if Auto Start is
enabled. Most FastCGI/LSAPI applications can only process one request
per process instance and for those type of applications, instances should
be set to match the value of Max Connections. Some FastCGI/LSAPI applications can spawn multiple child processes to
handle multiple requests concurrently and for these type of applications,
instances should be set to '1' and let the application spawn
Max Connections number of children processes.
|
| Syntax: Integer number |
|
| Run On Start Up | |
|
| Description: Specifies whether to start the external application at server start
up. Only applicable to external applications that can manage its own child processes and where 'Instances' value is set to '1'.
If enabled, external processes will be created at server startup instead of run-time.
|
| Syntax: Select from radio box |
| Tips: [Performance] If configured external process has significant startup overhead, such as Rail apps, then
this option should be enabled to decrease first-page response time. |
|
| Priority | |
|
| Description: Specifies priority of the external application process. Value ranges from
-20 to 20, lower number means higher priority. This
priority cannot be set higher (lower number) than server's own. Otherwise, server's
priority will be used and this value is disregarded.
|
| Syntax: int |
| See Also: Server Priority |
|
| Memory Soft Limit | |
|
| Description: Specifies memory consumption limit in bytes for an external application process
or an external application started by server.
The main purpose of this limit is to prevent excessive memory usage
because of software bug or intention attacks, not to impose a limit on
normal usage, make sure to leave enough head room, otherwise your
application will fail and 503 error may be returned. It can be set at server
level or at
individual external application level. Server level limit will be used if
it is not set at individual application level.
Operating system's default will be used if the value is set to
0 or absent at both levels.
|
| Syntax: Integer number |
| Tips: [Attention] Do not over tune this limit, it may result in 503 errors
if your application need more memory. |
|
| Memory Hard Limit | |
|
| Description: same as Memory Soft Limit, "Soft Limit" can be raised up to the
"Hard Limit" from within a user process. It can be set at server
level or at
individual external application level. Server level limit will be used if
it is not set at individual application level.
Operating system's default will be used if the value is set to
0 or absent at both levels.
|
| Syntax: Integer number |
|
| Process Soft Limit | |
|
Description: Limits the total number of processes can be created on behalf
of a user.
This effectively limits the number of processes can be launched by
web server on behalf of a user. All existing processes will be counted
against this limit, not just new process to be started.
If the limit is set to 10, and there are more than 10 processes
running under one user, then no new process can be started by web
server for that user (suEXEC).
The main purpose of this limit is to prevent "fork bomb"
attacks or excessive usage, not to impose a limit on normal usage,
make sure to leave enough head room. It can be set at server level or at
individual external application level. Server level limit will be used if
it is not set at individual application level. Operating system's
default value will be used if this value is 0 or absent at
both levels.
|
| Syntax: Integer number |
|
| Process Hard Limit | |
|
| Description: Same as Process Soft Limit, "Soft Limit" can be raised up to the
"Hard Limit" from within a user process. It can be set at server level or at
individual external application level. Server level limit will be used if
it is not set at individual application level.
Operating system's default value will be used if the value is set to
0 or absent at both levels.
|
| Syntax: Integer number |
|
| |
|
|
| |
Copyright © 2003-2007. Lite Speed Technologies Inc.
All rights reserved. |