Apache Module mod_proxy

NoteIn addition to the nonce, the |

Less powerful server, don't send as many requests there, BalancerMember ajp://1.2.3.6:8009 loadfactor=5

The server below is on hot standby BalancerMember ajp://1.2.3.6:8009 status=+H ProxySet lbmethod=bytraffic

Do not enable proxying with

until you have ProxyRequestssecured your server. Open proxy servers are dangerous both to your network and to the Internet at large.

and related modules implement a proxy/gateway for Apache HTTP Server, supporting a number of popular protocols as well as several different load balancing algorithms. Third-party modules can add support for additional protocols and load balancing algorithms.mod_proxy

A set of modules must be loaded into the server to provide the necessary features. These modules can be included statically at build time or dynamically via the

, which provides basic proxy capabilitiesmodproxybalancer

for more information.)| Protocol | Module | |---|---| | AJP13 (Apache JServe Protocol version 1.3) | | | CONNECT (for SSL) | | | FastCGI | | | ftp | | | HTTP/0.9, HTTP/1.0, and HTTP/1.1 | | | SCGI | | | WS and WSS (Web-sockets) | |

In addition, extended features are provided by other modules. Caching is provided by

and related modules. The ability to contact remote servers using the SSL/TLS protocol is provided by the mod_cacheSSLProxy*

. These additional modules will need to be loaded and configured to take advantage of these features.mod_ssl

An ordinary forward proxy is an intermediate server that sits between the client and the origin server. In order to get content from the origin server, the client sends a request to the proxy naming the origin server as the target and the proxy then requests the content from the origin server and returns it to the client. The client must be specially configured to use the forward proxy to access other sites.

A typical usage of a forward proxy is to provide Internet access to internal clients that are otherwise restricted by a firewall. The forward proxy can also use caching (as provided by

directive. Because forward proxies allow clients to access arbitrary sites through your server and to hide their true origin, it is essential that you ProxyRequestssecure your server so that only authorized clients can access the proxy before activating a forward proxy.

A reverse proxy (or gateway), by contrast, appears to the client just like an ordinary web server. No special configuration on the client is necessary. The client makes ordinary requests for content in the name-space of the reverse proxy. The reverse proxy then decides where to send those requests, and returns the content as if it was itself the origin.

A typical usage of a reverse proxy is to provide Internet users access to a server that is behind a firewall. Reverse proxies can also be used to balance load among several back-end servers, or to provide caching for a slower back-end server. In addition, reverse proxies can be used simply to bring several servers into the same URL space.

directive. It is RewriteRulenot necessary to turn

In addition, if you wish to have caching enabled, consult the documentation from

ProxyPass /foo http://foo.example.com/bar ProxyPassReverse /foo http://foo.example.com/bar

You can also force a request to be handled as a reverse-proxy request, by creating a suitable Handler pass-through. The example configuration below will pass all requests for PHP scripts to the specified FastCGI server using reverse proxy:

SetHandler "proxy:unix:/path/to/app.sock|fcgi://localhost/"

The two default workers have a fixed configuration and will be used if no other worker matches the request. They do not use HTTP Keep-Alive or connection pooling. The TCP connections to the origin server will instead be opened and closed for each request.

Explicitly configured workers are identified by their URL. They are usually created and configured using

Using explicitly configured workers in the forward mode is not very common, because forward proxies usually communicate with many different origin servers. Creating explicit workers for some of the origin servers can still be useful, if they are used very often. Explicitly configured workers have no concept of forward or reverse proxying by themselves. They encapsulate a common concept of communication with origin servers. A worker created by

for use in a reverse proxy will be also used for forward proxy requests whenever the URL to the origin server matches the worker URL and vice versa.ProxyPass

The URL identifying a direct worker is the URL of its origin server including any path components given:

ProxyPass /examples http://backend.example.com/examples ProxyPass /docs http://backend.example.com/docs

This example defines two different workers, each using a separate connection pool and configuration.

Worker sharing happens if the worker URLs overlap, which occurs when the URL of some worker is a leading substring of the URL of another worker defined later in the configuration file. In the following example

the second worker isn't actually created. Instead the first worker is used. The benefit is, that there is only one connection pool, so connections are more often reused. Note that all configuration attributes given explicitly for the later worker will be ignored. This will be logged as a warning. In the above example the resulting timeout value for the URL /examples

If you want to avoid worker sharing, sort your worker definitions by URL length, starting with the longest worker URLs. If you want to maximize worker sharing use the reverse sort order. See also the related warning about ordering

Explicitly configured workers come in two flavors: direct workers and (load) balancer workers. They support many important configuration attributes which are described below in the

The set of options available for a direct worker depends on the protocol, which is specified in the origin server URL. Available protocols include ajp

Balancer workers are virtual workers that use direct workers known as their members to actually handle the requests. Each balancer can have multiple members. When it handles a request, it chooses a member based on the configured load balancing algorithm.

as the protocol scheme. The balancer URL uniquely identifies the balancer worker. Members are added to a balancer using

Require ip 192.168.0

Strictly limiting access is essential if you are using a forward proxy (using the

directive). Otherwise, your server can be used by any client to access arbitrary hosts while hiding his or her true identity. This is dangerous both for your network and for the Internet at large. When using a reverse proxy (using the ProxyRequests

), access control is less critical because clients can only contact the hosts that you have specifically configured.

See Also the Proxy-Chain-Auth environment variable.

directive to forward the respective ProxyRemotescheme to the firewall proxy). However, when it has to access resources within the intranet, it can bypass the firewall when accessing hosts. The

directive is useful for specifying which hosts belong to the intranet and should be accessed directly.NoProxy

Users within an intranet tend to omit the local domain name from their WWW requests, thus requesting "http://somehost/" instead of http://somehost.example.com/

. Some commercial proxy servers let them get away with this and simply serve the request, implying a configured local domain. When the

directive is used and the server is ProxyDomainconfigured for proxy service, Apache httpd can return a redirect response and send the client to the correct, fully qualified, server address. This is the preferred method since the user's bookmark files will then contain fully qualified hosts.

ProxyPass http://buggyappserver:7001/foo/ SetEnv force-proxy-request-1.0 1 SetEnv proxy-nokeepalive 1

. But if the body is large and the original request used chunked encoding, then chunked encoding may also be used in the upstream request. You can control this selection using environment variables. Setting proxy-sendcl

ensures maximum compatibility with upstream servers by always sending the Content-Length

minimizes resource usage by using chunked encoding.

Under some circumstances, the server must spool request bodies to disk to satisfy the requested handling of request bodies. For example, this spooling will occur if the original body was sent with chunked encoding (and is large), but the administrator has asked for backend requests to be sent with Content-Length or as HTTP/1.0. This spooling can also occur if the request body already has a Content-Length header, but the server is configured to filter incoming request bodies.

only applies to request bodies that the server will spool to diskLimitRequestBody

Be careful when using these headers on the origin server, since they will contain more than one (comma-separated) value if the original request already contained one of these headers. For example, you can use %{X-Forwarded-For}i

in the log format string of the origin server to log the original clients IP address, but you may get more than one address if the request passes through several proxies.

Note: If you need to specify custom request headers to be added to the forwarded request, use the

The setting in the global server defines the default for all vhosts.

directives: loadfactor. This is the member load factor - a number between 1 (default) and 100, which defines the weighted load to be applied to the member in question.

The balancerurl is only needed when not in

Trailing slashes should typically be removed from the URL of a BalancerMember

A Domain is a partially qualified DNS domain name, preceded by a period. It represents a list of hosts which logically belong to the same DNS domain or zone (i.e., the suffixes of the hostnames are all ending in Domain).

To distinguish Domains from Hostnames (both syntactically and semantically; a DNS domain can have a DNS A record, too!),

Domain name comparisons are done without regard to the case, and Domains are always assumed to be anchored in the root of the DNS tree, therefore two domains .ExAmple.com

(note the trailing period) are considered equal. Since a domain comparison does not involve a DNS lookup, it is much more efficient than subnet comparison.

A SubNet is a partially qualified internet address in numeric (dotted quad) form, optionally followed by a slash and the netmask, specified as the number of significant bits in the SubNet. It is used to represent a subnet of hosts which can be reached over a common network interface. In the absence of the explicit net mask it is assumed that omitted (or zero valued) trailing digits specify the mask. (In this case, the netmask can only be multiples of 8 bits wide.) Examples:

255.255.0.0

)As a degenerate case, a SubNet with 32 valid bits is the equivalent to an IPAddr, while a

A IPAddr represents a fully qualified internet address in numeric (dotted quad) form. Usually, this address represents a host, but there need not necessarily be a DNS domain name connected with the address.

192.168.123.7

An IPAddr does not need to be resolved by the DNS system, so it can result in more effective apache performance.

A Hostname is a fully qualified DNS domain name which can be resolved to one or more IPAddrs via the DNS domain name service. It represents a logical host (in contrast to

prep.ai.example.edu

SetOutputFilter INCLUDES

A backend URL matches the configuration section if it begins with the the wildcard-url string, even if the last path segment in the directive only matches a prefix of the backend URL. For example, matches all of http://example.com/foo, http://example.com/foo/bar, and http://example.com/foobar. The matching of the final URL differs from the behavior of the

section, which for purposes of this note treats the final path component as if it ended in a slash.

ProxyBadHeader IsError|Ignore|StartBody

ProxyBlock *|word|host|domain [word|host|domain] ...

directive specifies a list of words, hosts and/or domains, separated by spaces. HTTP, HTTPS, and FTP document requests to sites whose names contain matched words, hosts or domains are blocked by the proxy server. The proxy module will also attempt to determine IP addresses of list items which may be hostnames during startup, and cache them for match test as well. That may slow down the startup time of the server.

ProxyBlock news.example.com auctions.example.com friends.example.com

directive specifies the default domain which the apache proxy server will belong to. If a request to a host without a domain name is encountered, a redirection response to the same host with the configured Domain appended will be generated.

's SSI) to get the error code and act accordingly (default behavior would display the error page of the proxied server, turning this on shows the SSI Error message).mod_include

This directive does not affect the processing of informational (1xx), normal success (2xx), or redirect (3xx) responses.

In almost every case there's no reason to change that value.

If used with AJP this directive sets the maximum AJP packet size in bytes. Values larger than 65536 are set to 65536. If you change it from the default, you must also change the packetSize

attribute of your AJP connector on the Tomcat side! The attribute packetSize

Normally it is not necessary to change the maximum packet size. Problems with the default value have been reported when sending certificates or certificate chains.

From 2.4.8 onwards, named groups and backreferences are captured and written to the environment with the corresponding name prefixed with "MATCH_" and in upper case. This allows elements of URLs to be referenced from within expressions and modules like

. In order to prevent confusion, numbered (unnamed) backreferences are ignored. Use named groups instead.mod_rewrite

[^/]+)> Require ldap-group cn=%{env:MATCH_SITENAME},ou=combined,o=Example

is a violation of the HTTP/1.1 protocol (RFC2616), which forbids a Proxy setting Max-Forwards

if the Client didn't set it. Earlier Apache httpd versions would always set it. A negative ProxyMaxForwards

value, including the default -1, gives you protocol-compliant behaviour, but may leave you open to loops.

ProxyPass [path] !|url [key=value [key=value ...]] [nocanon] [interpolate] [noquery]

This directive allows remote servers to be mapped into the space of the local server; the local server does not act as a proxy in the conventional sense, but appears to be a mirror of the remote server. The local server is often called a reverse proxy or gateway. The path is the name of a local virtual path; url is a partial URL for the remote server and cannot include a query string.

context.Support for using a Unix Domain Socket is available by using a target which prepends unix:/path/lis.sock|

. For example, to proxy HTTP and target the UDS at /home/www/socket you would use unix:/home/www.socket|http://localhost/whatever/

aware.Suppose the local server has address http://example.com/

to be internally converted into a proxy request to http://backend.example.com/bar

The following alternative syntax is possible, however it can carry a performance penalty when present in very large numbers. The advantage of the below syntax is that it allows for dynamic control via the Balancer Manager interface:

If the first argument ends with a trailing /, the second argument should also end with a trailing / and vice versa. Otherwise the resulting requests to the backend may miss some needed slashes and do not deliver the expected results.

rules are checked in the order of configuration. The first rule that matches wins. So usually you should sort conflicting ProxyPassMatch

rules starting with the longest URLs first. Otherwise later rules for longer URLS will be hidden by any earlier rule which uses a leading substring of the URL. Note that there is some relation with worker sharing. In contrast, only one ProxyPass

For the same reasons exclusions must come before the general ProxyPass

In Apache HTTP Server 2.1 and later, mod_proxy supports pooled connections to a backend server. Connections created on demand can be retained in a pool for future use. Limits on the pool size and other settings can be coded on the ProxyPass

By default, mod_proxy will allow and retain the maximum number of connections that could be used simultaneously by that web server child process. Use the max

parameter to reduce the number from the default. Use the ttl

parameter to set an optional time to live; connections which have been unused for at least ttl

can be used to avoid using a connection which is subject to closing because of the backend server's keep-alive timeout.

and other settings are not coordinated among all child processes, except when only one child process is allowed by configuration or MPM design.

| Parameter | Default | Description | |||||| |---|---|---|---|---|---|---|---|---| | min | 0 | Minimum number of connection pool entries, unrelated to the actual number of connections. This only needs to be modified from the default for special circumstances where heap memory associated with the backend connections should be preallocated or retained. | |||||| | max | 1...n | Maximum number of connections that will be allowed to the backend server. The default for this limit is the number of threads per process in the active MPM. In the Prefork MPM, this is always 1, while with other MPMs it is controlled by the ThreadsPerChild directive. | |||||| | smax | max | Retained connection pool entries above this limit are freed during certain operations if they have been unused for longer than the time to live, controlled by the ttl parameter. If the connection pool entry has an associated connection, it will be closed. This only needs to be modified from the default for special circumstances where connection pool entries and any associated connections which have exceeded the time to live need to be freed or closed more aggressively. | |||||| | acquire | - | If set this will be the maximum time to wait for a free connection in the connection pool, in milliseconds. If there are no free connections in the pool the Apache httpd will return SERVERBUSY status to the client. | |||||| | connectiontimeout | timeout | Connect timeout in seconds. The number of seconds Apache httpd waits for the creation of a connection to the backend to complete. By adding a postfix of ms the timeout can be also set in milliseconds. | |||||| | disablereuse | Off | This parameter should be used when you want to force modproxy to immediately close a connection to the backend after being used, and thus, disable its persistent connection and pool for that backend. This helps in various situations where a firewall between Apache httpd and the backend server (regardless of protocol) tends to silently drop connections or when backends themselves may be under round- robin DNS. To disable connection pooling reuse, set this property value to On . | |||||| | flushpackets | off | Determines whether the proxy module will auto-flush the output brigade after each "chunk" of data. 'off' means that it will flush only when needed, 'on' means after each chunk is sent and 'auto' means poll/wait for a period of time and flush if no input has been received for 'flushwait' milliseconds. Currently this is in effect only for AJP. | |||||| | flushwait | 10 | The time to wait for additional input, in milliseconds, before flushing the output brigade if 'flushpackets' is 'auto'. | |||||| | iobuffersize | 8192 | Adjusts the size of the internal scratchpad IO buffer. This allows you to override the ProxyIOBufferSize for a specific worker. This must be at least 512 or set to 0 for the system default of 8192. | |||||| | keepalive | Off | This parameter should be used when you have a firewall between your Apache httpd and the backend server, who tend to drop inactive connections. This flag will tell the Operating System to send The frequency of initial and subsequent TCP keepalive probes depends on global OS settings, and may be as high as 2 hours. To be useful, the frequency configured in the OS must be smaller than the threshold used by the firewall. | |||||| | lbset | 0 | Sets the load balancer cluster set that the worker is a member of. The load balancer will try all members of a lower numbered lbset before trying higher numbered ones. | |||||| | ping | 0 | Ping property tells the webserver to "test" the connection to the backend before forwarding the request. For AJP, it causes to send a CPING request on the ajp13 connection (implemented on Tomcat 3.3.2+, 4.1.28+ and 5.0.13+). For HTTP, it causes to send a 100-Continue to the backend (only valid for HTTP/1.1 - for non HTTP/1.1 backends, this property has no effect). In both cases the parameter is the delay in seconds to wait for the reply. This feature has been added to avoid problems with hung and busy backends. This will increase the network traffic during the normal operation which could be an issue, but it will lower the traffic in case some of the cluster nodes are down or busy. By adding a postfix of ms the delay can be also set in milliseconds. | |||||| | receivebuffersize | 0 | Adjusts the size of the explicit (TCP/IP) network buffer size for proxied connections. This allows you to override the ProxyReceiveBufferSize for a specific worker. This must be at least 512 or set to 0 for the system default. | |||||| | redirect | - | Redirection Route of the worker. This value is usually set dynamically to enable safe removal of the node from the cluster. If set all requests without session id will be redirected to the BalancerMember that has route parameter equal as this value. | |||||| | retry | 60 | Connection pool worker retry timeout in seconds. If the connection pool worker to the backend server is in the error state, Apache httpd will not forward any requests to that server until the timeout expires. This enables to shut down the backend server for maintenance, and bring it back online later. A value of 0 means always retry workers in an error state with no timeout. | |||||| | route | - | Route of the worker when used inside load balancer. The route is a value appended to session id. | |||||| | status | - | Single letter value defining the initial status of this worker. | |||||| | timeout | | Connection timeout in seconds. The number of seconds Apache httpd waits for data sent by / to the backend. | |||||| | ttl | - | Time to live for inactive connections and associated connection pool entries, in seconds. Once reaching this limit, a connection will not be used again; it will be closed at some later time. |

, any path information is ignored) then a virtual worker that does not really communicate with the backend server will be created. Instead it is responsible for the management of several "real" workers. In that case the special set of parameters can be add to this virtual worker. See

for more information about how the balancer works. modproxybalancer

| Parameter | Default | Description | |---|---|---| | lbmethod | byrequests | Balancer load-balance method. Select the load-balancing scheduler method to use. Either byrequests , to perform weighted request counting, bytraffic , to perform weighted traffic byte count balancing, or bybusyness , to perform pending request balancing. Default is byrequests . | | maxattempts | One less than the number of workers, or 1 with a single worker. | Maximum number of failover attempts before giving up. | | nofailover | Off | If set to On the session will break if the worker is in error state or disabled. Set this value to On if backend servers do not support session replication. | | stickysession | - | Balancer sticky session name. The value is usually set to something like JSESSIONID or PHPSESSIONID , and it depends on the backend application server that support sessions. If the backend application server uses different name for cookies and url encoded id (like servlet containers) use | to to separate them. The first part is for the cookie the second for the path.Available in Apache HTTP Server 2.4.4 and later. | | stickysessionsep | "." | Sets the separation symbol in the session cookie. Some backend application servers do not use the '.' as the symbol. For example the Oracle Weblogic server uses '!'. The correct symbol can be set using this option. The setting of 'Off' signifies that no symbol is used. | | scolonpathdelim | Off | If set to On the semi-colon character ';' will be used as an additional sticky session path delimiter/separator. This is mainly used to emulate mod_jk's behavior when dealing with paths such as JSESSIONID=6736bcf34;foo=aabfa | | timeout | 0 | Balancer timeout in seconds. If set this will be the maximum time to wait for a free worker. Default is not to wait. | | failonstatus | - | A single or comma-separated list of HTTP status codes. If set this will force the worker into error state when the backend returns any status code in the list. Worker recovery behaves the same as other worker errors. | | failontimeout | Off | If set, an IO read timeout after a request is sent to the backend will force the worker into error state. Worker recovery behaves the same as other worker errors. Available in Apache HTTP Server 2.4.5 and later. | | nonce | | The protective nonce used in the balancer-manager application page. The default is to use an automatically determined UUID-based nonce, to provide for further protection for the page. If set, then the nonce is set to that value. A setting of None disables all nonce checking. ## NoteIn addition to the nonce, the | | growth | 0 | Number of additional BalancerMembers to allow to be added to this balancer in addition to those defined at configuration. | | forcerecovery | On | Force the immediate recovery of all workers without considering the retry parameter of the workers if all workers of a balancer are in error state. There might be cases where an already overloaded backend can get into deeper trouble if the recovery of all workers is enforced without considering the retry parameter of each worker. In this case set to Off .Available in Apache HTTP Server 2.4.2 and later. |

A sample balancer setup

ProxyPass /special-area http://special.example.com smax=5 max=10 ProxyPass / balancer://mycluster/ stickysession=JSESSIONID|jsessionid nofailover=On BalancerMember ajp://1.2.3.4:8009 BalancerMember ajp://1.2.3.5:8009 loadfactor=20 # Less powerful server, don't send as many requests there, BalancerMember ajp://1.2.3.6:8009 loadfactor=5

ProxyPass / balancer://hotcluster/ BalancerMember ajp://1.2.3.4:8009 loadfactor=1 BalancerMember ajp://1.2.3.5:8009 loadfactor=2 # The server below is on hot standby BalancerMember ajp://1.2.3.6:8009 status=+H ProxySet lbmethod=bytraffic

Normally, modproxy will canonicalise ProxyPassed URLs. But this may be incompatible with some backends, particularly those that make use of PATHINFO. The optional nocanon keyword suppresses this, and passes the URL path "raw" to the backend. Note that may affect the security of your backend, as it removes the normal limited protection against URL-based attacks provided by the proxy.

Normally, modproxy will include the query string when generating the SCRIPTFILENAME environment variable. The optional noquery keyword (available in httpd 2.4.1 and later) prevents this.

section, the first argument is omitted and the local directory is obtained from the

section, however ProxyPass does not interpret the regexp as such, so it is necessary to use ProxyPassMatch

directive with the RewriteRule[P]

causes the ProxyPass to interpolate environment variables, using the syntax ${VARNAME}. Note that many of the standard CGI-derived environment variables will not exist when this interpolation happens, so you may still have to resort to

for complex rules. Also note that interpolation is not supported within the scheme portion of a URL. Dynamic determination of the scheme can be accomplished with mod_rewrite

RewriteEngine On RewriteCond %{HTTPS} =off RewriteRule . - [E=protocol:http] RewriteCond %{HTTPS} =on RewriteRule . - [E=protocol:https] RewriteRule ^/mirror/foo/(.*) %{ENV:protocol}://backend.example.com/$1 [P] ProxyPassReverse /mirror/foo/ http://backend.example.com/ ProxyPassReverse /mirror/foo/ https://backend.example.com/

Keep this turned off (for server performance) unless you need it!

The URL argument must be parsable as a URL before regexp substitutions (as well as after). This limits the matches you can use. For instance, if we had used

in our previous example, it would fail with a syntax error at server startup. This is a bug (PR 46665 in the ASF bugzilla), and the workaround is to reformulate the match:

Take care when constructing the target URL of the rule, considering the security impact from allowing the client influence over the set of URLs to which your server will act as a proxy. Ensure that the scheme and hostname part of the URL is either fixed, or does not allow the client undue influence.

Only the HTTP response headers specifically mentioned above will be rewritten. Apache httpd will not rewrite other response headers, nor will it by default rewrite URL references inside HTML pages. This means that if the proxied content contains absolute URL references, they will by-pass the proxy. To rewrite HTML content to match the proxy, you must load and enable

before forwarding the HTTP redirect response to the client. Note that the hostname used for constructing the URL is chosen in respect to the setting of the

section, but will probably not work as intended, as ProxyPassReverse will interpret the regexp literally as a path; if needed in this situation, specify the ProxyPassReverse outside the section, or in a separate

or, in fact, anything) to /mirror/foo/

. It is mostly useful in special configurations like proxied mass name-based virtual hosting, where the original Host header needs to be evaluated by the backend server.

ProxyRemote http://goodguys.example.com/ http://mirrorguys.example.com:8000 ProxyRemote * http://cleverproxy.localdomain ProxyRemote ftp http://ftpproxy.mydomain:8080

container directive, the balancer url|worker url>url argument is not required. As a side effect the respective balancer or worker gets created. This can be useful when doing reverse proxying via a

BalancerMember http://www2.example.com:8080 loadfactor=1 BalancerMember http://www3.example.com:8080 loadfactor=2 ProxySet lbmethod=bytraffic

Keep in mind that the same parameter key can have a different meaning depending whether it is applied to a balancer or a worker as shown by the two examples above regarding timeout.

Full is synonymous with On

Modules | Directives | FAQ | Glossary | Sitemap

Apache HTTP Server Version 2.4