pound.8

.TH POUND "8" "Jan 2010" "pound" "System Manager's Manual"
.SH NAME
pound \- HTTP/HTTPS reverse-proxy and load-balancer
.SH SYNOPSIS
.TP
.B pound
[\fI-v\fR]
[\fI-c\fR]
[\fI-W\fR]
[\fI-V\fR]
[\fI-f config_file\fR]
[\fI-p pid_file\fR]
.SH DESCRIPTION
.PP
.B Pound
is a reverse-proxy load balancing server. It accepts requests from HTTP/HTTPS
clients and distributes them to one or more Web servers. The HTTPS requests are
decrypted and passed to the back-ends as plain HTTP.
.PP
If more than one back-end server is defined,
.B Pound
chooses one of them randomly, based on defined priorities. By default,
.B Pound
keeps track of associations between clients and back-end servers (sessions).
.SH GENERAL PRINCIPLES
.P
In general
.B Pound
needs three types of objects defined in order to function:
.IR listeners ,
.I services
and
.IR back-ends .
.TP
\fBListeners\fR
A
.I listener
is a definition of how
.B Pound
receives requests from the clients (browsers). Two types of
.I listeners
may be defined: regular HTTP
.I listeners
and HTTPS (HTTP over SSL/TLS)
.IR listeners .
At the very least a
.I listener
must define the address and port to listen on, with additional
requirements for HTTPS
.IR listeners .
.TP
\fBServices\fR
A
.I service
is the definition of how the requests are answered. The
.I services
may be defined within a
.I listener
or at the top level (global). When a request is received
.B Pound
attempts to match them to each
.I service
in turn, starting with the
.I services
defined in the
.I listener
itself and, if needed, continuing with the
.I services
defined at the global level. The
.I services
may define their own conditions as to which requests they can answer:
typically this involves certain URLs (images only, or a certain path)
or specific headers (such as the Host header). A
.I service
may also define a
.I session
mechanism: if defined future requests from a given client will always
be answered by the same
.IR back-end .
.TP
\fBBack-ends\fR
The
.I back-ends
are the actual servers for the content requested. By itself,
.B Pound
supplies no responses - all contents must be received from a "real"
web server. The
.I back-end
defines how the server should be contacted.
.IP
Three types of
.I back-ends
may be defined: a "regular"
.I back-end
which receives requests and returns responses, a "redirect"
.I back-end
in which case
.B Pound
will respond with a redirect response, without accessing any
.I back-end
at all, or an "emergency"
.I back-end
which will be used only if all other backends are "dead".
.IP
Multiple
.I back-ends
may be defined within a
.IR service ,
in which case
.B Pound
will load-balance between the available
.IR back-ends .
.IP
If a
.I back-end
fails to respond it will be considered "dead", in which case
.B Pound
will stop sending requests to it. Dead
.I back-ends
are periodically checked for availability, and once they respond again they
are "resurected" and requests are sent again their way. If no
.I back-ends
are available (none were defined, or all are "dead") then
.B Pound
will reply with "503 Service Unavailable", without checking additional
.IR services .
.IP
The connection between
.B Pound
and the
.I back-ends
is always via HTTP, regardless of the actual protocol used between
.B Pound
and the client.
.SH OPTIONS
Options available (see also below for configuration file options):
.TP
\fB\-v\fR
Verbose mode: error messages will be sent to stdout even if
.B Pound
was configured to log to syslog. This applies only to startup messages, before
.B Pound
puts itself in the background. Normal operational messages will still go to syslog.
.TP
\fB\-V\fR
Print version:
.B Pound
will exit immediately after printing the current version and configuration flags.
.TP
\fB\-c\fR
Check only:
.B Pound
will exit immediately after parsing the configuration file. This may be used for
running a quick syntax check before actually activating a server.
.TP
\fB\-W\fR
Check WAF configuration:
.B Pound
will exit immediately after parsing the WAF configuration rules 'WafRules'. This
is used to check if the WAF rules are well built and if a conflict exists among
the imported rulesets.
.TP
\fB\-f\fR config_file
Location of the configuration file (see below for a full description of the format).
Default:
.I /usr/local/etc/pound.cfg
.TP
\fB\-p\fR pid_file
Location of the pid file.
.B Pound
will write its own pid into this file. Normally this is used for shell
scripts that control starting and stopping of the daemon.
Default:
.I /var/run/pound.pid
.PP
In general, any number of back-end servers may be specified. Use the priority to
affect the load distribution among unequal-performance servers.
.PP
One (or more) copies of
.B Pound
should be started at boot time. Use "big iron" if you expect heavy loads: while
.B Pound
is as light-weight as I know how to make it, with a lot of simultaneous requests it
will use quite a bit of CPU and memory. Multiple CPUs are your friend.
.SH "CONFIGURATION FILE"
Each line in the file is considered a complete configuration directive. The directives
are case-insensitive. Empty lines or lines starting in '#' are ignored. There are three
types of directives:
.B global
directives (they affect the settings for the entire program instance),
.B listener
directives (they define which requests
.B Pound
will listen for), and
.B service
directives (they affect only a specific group of requests).
.SH "GLOBAL DIRECTIVES"
Global directives may appear anywhere within the configuration file, though it is
customary for them to be at the start. They may appear in any order.
.TP
\fBUser\fR "user_name"
Specify the user
.B Pound
will run as (must be defined in \fI/etc/passwd\fR).
.TP
\fBGroup\fR "group_name"
Specify the group
.B Pound
will run as (must be defined in \fI/etc/group\fR).
.TP
\fBName\fR SortName
Specify a Sort name without blank spaces for the Process, this information will shown in logs
.TP
\fBRootJail\fR "directory_path_and_name"
Specify the directory that
.B Pound
will chroot to at runtime. Please note that OpenSSL requires access to /dev/urandom,
so make sure you create a device by that name, accessible from the root jail
directory.
.B Pound
may also require access to
.I /dev/syslog
or similar.
.TP
\fBDaemon\fR 0|1
Have
.B Pound
run in the foreground (if 0) or as a daemon (if 1). By default
.B Pound
runs as a daemon (detaches itself from the controlling terminal and
puts itself in the background). By specifying this option you can force
.B Pound
to work like a regular process. Useful for debugging or if you want to
use something like \fIdaemontools\fR.
.TP
\fBThreadModel\fR [pool|dynamic]
If set to pool,
.B Pound
will spawn
\fBThreads\fR
number of threads when it starts and reuse those threads throughout its
lifetime.  This sets a hard limit on the number of concurrent requests
that can be processed.
If set to dynamic, a new thread is spawned for each and every request.
This means you have a penalty for thread creation on every request, and
possibility of DoS if too many requests come in, but it also means your
thread count will scale up and down as necessary.
.TP
\fBThreads\fR nnn
How many worker threads
.B Pound
should use. Default: 128. Tune this parameter to improve performance.
If you set it too high,
.B Pound
will use a lot memory, and some CPU will be wasted on context switches.
If you set it too low requests may be served with some delay. Experiment
to find the optimal value for your installation.
.TP
\fBLogFacility\fR value
Specify the log facility to use.
.I value
(default: daemon) must be one of the symbolic facility names defined in
\fIsyslog.h\fR. This facility shall be used for logging. Using a - for
the facility name causes
.B Pound
to log to stdout/stderr.
.TP
\fBDHParams\fR "path/to/dhparams.pem"
Use the supplied dhparams pem file for DH key exchange for non-export-controlled
negotiations.  Generate such a file with \fBopenssl dhparam\fR.
This can be used to do 2048bit DHE.
.TP
\fBLogLevel\fR value
Specify the logging level: 0 for no logging, 1 (default) for regular
logging, 2 for extended logging (show chosen backend server as well),
3 for Apache-like format (Combined Log Format with Virtual Host), 4
(same as 3 but without the virtual host information) and 5 (same as 4
but with information about the
.I Service
and
.I BackEnd
used).
This value can be overridden for specific listeners.
.TP
\fBIgnoreCase\fR 0|1
Ignore case when matching URLs (default: 0). This value can be
overridden for specific services.
.TP
\fBIgnore100continue\fR 0|1
Ignore Header Expect: 100-continue (default: 1, Ignored).
If 0 pound manages Expect: 100-continue headers.
.TP
\fBDynScale\fR 0|1
Enable or disable the dynamic rescaling code (default: 0). If enabled
.B Pound
will periodically try to modify the back-end priorities in order to
equalise the response times from the various back-ends.
This value can be overridden for specific services.
.TP
\fBAlive\fR value
Specify how often
.B Pound
will check for resurected back-end hosts (default: 30 seconds). In
general, it is a good idea to set this as low as possible - it
will find resurected hosts faster. However, if you set it too
low it will consume resources - so beware.
.TP
\fBClient\fR value
Specify for how long
.B Pound
will wait for a client request (default: 10 seconds). After this
long has passed without the client sending any data
.B Pound
will close the connection. Set it higher if your clients
time-out on a slow network or over-loaded server, lower if you
start getting DOS attacks or run into problems with IE clients.
This value can be overridden for specific listeners.
.TP
\fBTimeOut\fR value
How long should
.B Pound
wait for a response from the back-end (in seconds). Default: 15 seconds.
This value can be overridden for specific back-ends.
.TP
\fBConnTO\fR value
How long should
.B Pound
wait for a connection to the back-end (in seconds). Default: the
.B TimeOut
value. This value can be overridden for specific back-ends.
.TP
\fBWSTimeOut\fR value
How long should
.B Pound
wait for data from either back-end or client in a connection upgraded to
a WebSocket (in seconds). Default: 600 seconds.
This value can be overridden for specific back-ends.
.TP
\fBGrace\fR value
How long should
.B Pound
continue to answer existing connections after a receiving and INT or HUP
signal (default: 30 seconds). The configured listeners are closed
immediately. You can bypass this behaviour by stopping
.B Pound
with a TERM or QUIT signal, in which case the program exits without any
delay.
.TP
\fBSSLEngine\fR "name"
Use an OpenSSL hardware acceleration card called \fIname\fR. Available
only if OpenSSL-engine is installed on your system.
.TP
\fBECDHcurve\fR "name"
Use the named curve for elliptical curve encryption (default: prime256v1).
.TP
\fBControl\fR "/path/to/socket"
Set the control socket path. If not defined
.B Pound
does not listen for any commands. The commands may be issued by using
the
.I poundctl(8)
program.
.TP
\fBControlUser\fR "user"
The username to chown the Control socket to.
.TP
\fBControlGroup\fR "group"
The groupname to chgrp the Control socket to.
.TP
\fBControlMode\fR 0660
The mode the Control socket should use, in octal.
.TP
\fBInclude\fR "/path/to/file"
Include the file as though it were part of the configuration file.
.TP
\fBAnonymise\fR
Replace the last byte of the client address with 0 for logging purposes.
Default: log the client address in full.
\fBIncludeDir\fR "/path/"
Looks for files with .conf or .cfg extensions in "path", and includes all files, in sorted
order, inline in the configuration as if it were part of the configuration file.
This directive can be used in any block... but the result must be syntactically correct.
.TP
.SH "HTTP Listener"
An HTTP listener defines an address and port that
.B Pound
will listen on for HTTP requests. All configuration directives enclosed
between
.I ListenHTTP
and
.I End
are specific to a single HTTP listener. At the very least you must specify
and address and a port for each listener. The following directives are
available:
.TP
\fBAddress\fR address
The address that
.B Pound
will listen on. This can be a numeric IP address, or a symbolic host name
that must be resolvable at run-time.  This is a
.B mandatory
parameter. The address 0.0.0.0 may be used as an alias for 'all available
addresses on this machine', but this practice is strongly discouraged, as
it will interfere with the rewriting mechanisms (see below).
.TP
\fBPort\fR port
The port number that
.B Pound
will listen on.  This is a
.B mandatory
parameter.
.TP
\fBKey\fR "key"
The key associated to this backend, if using BackendCookie in the service.
If left blank, it'll be autogenerated from the backend address.
.TP
\fBxHTTP\fR value
Defines which HTTP verbs are accepted. The possible values are:
.IP
.I 0
(default) accept only standard HTTP requests (GET, POST, HEAD).
.IP
.I 1
additionally allow extended HTTP requests (PUT, PATCH, DELETE).
.IP
.I 2
additionally allow standard WebDAV verbs (LOCK, UNLOCK, PROPFIND,
PROPPATCH, SEARCH, MKCOL, MOVE, COPY, OPTIONS, TRACE, MKACTIVITY,
CHECKOUT, MERGE, REPORT).
.IP
.I 3
additionally allow MS extensions WebDAV verbs (SUBSCRIBE, UNSUBSCRIBE,
NOTIFY, BPROPFIND, BPROPPATCH, POLL, BMOVE, BCOPY, BDELETE, CONNECT).
.IP
.I 4
additionally allow MS RPC extensions verbs (RPC_IN_DATA, RPC_OUT_DATA).
.TP
\fBClient\fR value
Override the global
.I Client
time-out value.
.TP
\fBCheckURL\fR "pattern to match"
Define a pattern that must be matched by each request sent to this
listener. A request that does not match is considered to be illegal.
By default
.B Pound
accepts all requests (i.e. the pattern is ".*"), but you are free to
limit it to something more reasonable. Please note that this applies
only to the request path -
.B Pound
will still check that the request is syntactically correct.
.TP
\fBErr414\fR "filename"
A file with the text to be displayed if an Error 414 occurs.
Default: "Request URI is too long.".
.TP
\fBErr500\fR "filename"
A file with the text to be displayed if an Error 500 occurs.
Default: "An internal server error occurred. Please try again later.".
.TP
\fBErr501\fR "filename"
A file with the text to be displayed if an Error 501 occurs.
Default: "This method may not be used.".
.TP
\fBErr503\fR "filename"
A file with the text to be displayed if an Error 503 occurs.
Default: "The service is not available. Please try again later.".
.TP
\fBErrNoSsl\fR "filename"
A file with the text to be displayed if a user connects to a HTTPS listener with HTTP.
Default: "Please use HTTPS.".
.TP
\fBWafRule\fR "filename"
A file with SecLang directives to dectect and protect against HTTP malicious requests.
Several WafRule directives can be set, they will be analyzed in the same order than appears in
the pound configuration file.
\BWafBodySize\fR "bytes"
It is the buffer size, in bytes, for saving the request and response bodies. If the body is bigger than
this buffer, the body will be forwarded without being analyzed. If it has the value 0, the bodies will be
skipped. If it has the value -1 no limit is set for the buffer, the buffer will have the same size that
is set in the Content-Length HTTP header.
Default: "-1".

Only valid for HTTPS listeners.
.TP
\fBNoSslRedirect\fR [code] "url"
A url that the user will be redirected to if the user connects to a HTTPS listener with HTTP.
.br
The code here is just like the code in Redirect blocks. It defaults to 302, but could be 301 or 307.
.br
Only valid for HTTPS listeners.
.br
Example:
.IP
.br
NoSslRedirect "https://thishost:port"
.TP
\fBMaxRequest\fR nnn
Request maximal size. All requests will be limited to these many bytes. If
a request contains more data than allowed an error 414 is returned. Default:
unlimited.
.TP
\fBHeadRemove\fR "header pattern"
Remove certain headers from the incoming requests. All occurences of the
matching specified header will be removed. Please note that this filtering
is done prior to other checks (such as \fIHeadRequire\fR or \fIHeadDeny\fR),
so you should not try to check for these headers in later matches. Multiple
directives may be specified in order to remove more than one header, and
the header itself may be a regular pattern (though this should be used with
caution).
.TP
\RemoveResponseHead\fR "header pattern"
Remove certain headers from the outcomming response, the header sent by the
backend is not sent to the client. All occurences of the
matching specified header will be removed. Multiple directives may be specified
in order to remove more than one header, and the header itself may be a regular
pattern (though this should be used with caution).
.TP
\fBAddHeader\fR "header: to add"
Add the defined header to the request passed to the back-end server. The header
is added verbatim. Use multiple \fIAddHeader\fR directives if you need to add more
than one header.
.TP
\fBAddResponseHeader\fR "header: to add"
Add the defined header to the response passed to the client. The header
is added verbatim. Use multiple \fIAddHeader\fR directives if you need to add more
than one header.
.TP
\fBRewriteLocation\fR 0|1|2
If 1 force
.B Pound
to change the Location: and Content-location: headers in responses. If they
point to the back-end itself or to the listener (but with the wrong protocol)
the response will be changed to show the virtual host in the request. Default:
1 (active).  If the value is set to 2 only the back-end address is compared;
this is useful for redirecting a request to an HTTPS listener on
the same server as the HTTP listener.
.TP
\fBRewriteDestination\fR 0|1
If 1 force
.B Pound
to change the Destination: header in requests. The header is changed to point
to the back-end itself with the correct protocol. Default: 0.
.TP
\fBLogLevel\fR value
Override the global
.I LogLevel
value.
.TP
\fBForceHTTP10\fR "user agent pattern"
Force connections for browser user agents matching this pattern to use
HTTP/1.0 even if the browser has requested HTTP/1.1.  Some MSIE browsers have
problems with HTTP/1.1 over SSL connections.  Other browsers, like Java/1.0 or
JDK connections might require this feature as well.
.TP
\fBService\fR [ "name" ]
This defines a private service (see below for service definition syntax). This
service will be used only by this listener. The service may be optionally
named, with the name showing in the
.I poundctl
listings.
.SH "HTTPS Listener"
An HTTPS listener defines an address and port that
.B Pound
will listen on for HTTPS requests. All configuration directives enclosed
between
.I ListenHTTPS
and
.I End
are specific to a single HTTPS listener. At the very least you must specify
and address, a port and a server certificate for each listener. All directives
defined for HTTP listeners are applicable to HTTPS listeners as well. The
following additional directives are also available:
.TP
\fBCert\fR "certificate file"
Specify the server certificate. The
.I certificate file
is the file containing the certificate, possibly a certificate chain and the signature
for this server. This directive or the
.I CertDir
directive is
.B mandatory
for HTTPS listeners.
.IP
Please note that multiple
.I Cert
or
.I CertDir
directives are allowed if your OpenSSL version supports SNI. In such cases,
the first directive is the default certificate, with additional certificates
used if the client requests them.
.IP
The ordering of the directives is important: the first certificate where the CN
matches the client request will be used, so put your directives in the
most-specific-to-least specific order (i.e. wildcard certificates
.B after
host-specific certificates).
.IP
.I Cert
and
.I CertDir
directives
.B must
precede all other SSL-specific directives.
.TP
\fBCertDir\fR "certificate directory"
Specify the server certificate or certificates. The
.I certificate directory
is a directory path containing one or more certificates, possibly a certificate chain and the signature
for this server. This directive or
.I Cert
is
.B mandatory
for HTTPS listeners.
.IP
If a wildcard is specified, it will be honored.  Otherwise all files will be loaded from that directory.
For example, "/etc/certs/*.pem" will load all files from that directory that match the file extension given.
.IP
Please note that multiple
.I Cert
or
.I CertDir
directives are allowed if your OpenSSL version supports SNI. In such cases,
the first directive is the default certificate, with additional certificates
used if the client requests them.
.IP
The filenames in the directory will be sorted before being loaded. The order of files
is important: the first certificate where the CN
matches the client request will be used, so sort your files in the
most-specific-to-least specific order (i.e. wildcard certificates
.B after
host-specific certificates).
.IP
.I Cert
and
.I CertDir
directives
.B must
precede all other SSL-specific directives.
.TP
\fBClientCert\fR 0|1|2|3 depth
Ask for the client's HTTPS certificate: 0 - don't ask (default), 1 - ask,
2 - ask and fail if no certificate was presented, 3 - ask but do not verify.
.I Depth
is the depth of verification for a client certificate (up to 9). The default
depth limit is 9, allowing for the peer certificate and additional 9 CA
certificates that must be verified.
.TP
\fBDisable\fR SSLv2|SSLv3|TLSv1|TLSv1_1|TLSv1_2
Disable the protocol \fBand all lower protocols as well\fR.
This is due to a limitation in OpenSSL, which does not support disabling a single
protocol. For example,
.I Disable TLSv1
would disable SSLv2, SSLv3 and TLSv1, thus allowing only TLSv1_1 and TLSv1_2.
.TP
\fBCiphers\fR "acceptable:cipher:list"
This is the list of ciphers that will be accepted by the SSL connection; it is a
string in the same format as in OpenSSL
.I ciphers(1)
and
.I SSL_CTX_set_cipher_list(3).
.TP
\fBSSLHonorCipherOrder\fR 0|1
If this value is 1, the server will broadcast a preference to use \fBCiphers\fR in
the order supplied in the \fBCiphers\fR directive.  If the value is 0, the server
will treat the Ciphers list as the list of Ciphers it will accept, but no preference
will be indicated.  Default value is 0.
.TP
\fBSSLAllowClientRenegotiation\fR 0|1|2
If this value is 0, client initiated renegotiation will be disabled.  This will
mitigate DoS exploits based on client renegotiation, regardless of the patch status
of clients and servers related to "Secure renegotiation".  If the value is 1, secure
renegotiation is supported.  If the value is 2, insecure renegotiation is supported,
with unpatched clients. \fBThis can lead to a DoS and a Man in the Middle attack!\fR
The default value is 0.
.TP
\fBCAlist\fR "CAcert_file"
Set the list of "trusted" CA's for this server. The CAcert_file is a file containing
a sequence of CA certificates (PEM format). The names of the defined CA certificates
will be sent to the client on connection.
.TP
\fBVerifyList\fR "Verify_file"
Set the CA (Certificate Authority). The Verify_file is a file that contains the CA
root certificates (in PEM format).
.IP
.IR "Please note":
there is an important difference between the CAlist and the VerifyList. The
CAlist tells the client (browser) which client certificates it should send. The
VerifyList defines which CAs are actually used for the verification of the
returned certificate.
.TP
\fBCRLlist\fR "CRL_file"
Set the CRL (Certificate Revocation List) file. The CRL_file is a file that contains
the CRLs (in PEM format).
.TP
\fBNoHTTPS11\fR 0|1|2
Behave like an HTTP/1.0 server for HTTPS clients. If this value is
0 disable the check. If the value is 1 do not allow multiple
requests on SSL connections. If the value is 2 (default) disable multiple
requests on SSL connections only for MSIE clients. Required
work-around for a bug in certain versions of IE.
.TP
\fBSSLUncleanShutdown\fR "UserAgent Match Pattern"
Implement a workaround for MSIE's ssl shutdown code... basically don't
send a shutdown message because MSIE can't handle it.  Pound, by default,
would do this on all connections, and will do that if this directive is
never specified.  If this directive is specified, one or many times, any
matching useragent pattern will cause the ssl unclean shutdown behavior.
Other user agents will follow the mod_ssl compliant and safe code.
You likely want to set this for all MSIE browsers.
.TP
\fBForwardSNI\fR "0|1 default=1"
Enable SNI server host name forwarding to https backends if it presented by client.
.SH "Service"
.SH "Service"
A service is a definition of which back-end servers
.B Pound
will use to reply to incoming requests. A service may be defined as part
of a listener (in which case it will be used only by that listener), or
globally (which makes it available to all listeners).
.B Pound
will always try the private services in the order defined, followed by
the global ones.
.P
All configuration directives enclosed between
.I Service
and
.I End
are specific to a single service. The following directives are available:
.TP
\fBURL\fR "pattern"
Match the incoming request. If a request fails to match than this service
will be skipped and next one tried. If all services fail to match
.B Pound
returns an error. You may define multiple
.I URL
conditions per service, in which case
.B all
patterns must match. If no
.I URL
was defined then all requests match. The matching is by default case-sensitive,
but this can be overridden by specifying
.B IgnoreCase 1
.TP
\fBOrURLs\fR
Defines a block of
.I URL
directives that should be merged into a single pattern, all OR'd together.
This creates a pattern like
.B ((url1)|(url2)|(url3))
 for as many
.I URL
directives as are specified within the block.  End the block with an
.I End
directive.
.TP
\fBBackendCookie\fR "cookiename" "domain" "path" age|Session
If defined, Pound will inject a cookie in each response with the appropriate backend's key, so that
even if the session table is flushed or sessions are disabled, the proper backend can be chosen.
This allows for session databases to be offloaded to the client side via browser cookies.
See \fBKey\fR in the backend definition.  The given age will be how many seconds the cookie will
persist for.  If set to 0, it will be a so-called "memory" cookie which will expire when the browser
closes.  If set to "Session", it will mimick the session TTL behavior.
.TP
\fBIgnoreCase\fR 0|1
Override the global
.B IgnoreCase
setting.
.TP
\fBHeadRequire\fR "pattern"
The request must contain at least on header matching the given pattern.
Multiple
.I HeadRequire
directives may be defined per service, in which case all of them must
be satisfied.
.TP
\fBHeadDeny\fR "pattern"
The request may
.B not
contain any header matching the given pattern.  Multiple
.I HeadDeny
directives may be defined per service, in which case all of them must be satisfied.
.IP
.IR "Please note":
if the listener defined a
.I HeadRemove
directive, the matching headers are removed
.B before
the service matching is attempted.
.TP
\fBDynScale\fR 0|1
Enable or disable dynamic rescaling for the current service. This value will
override the value globally defined.
.TP
\fBDisabled\fR 0|1
Start
.B Pound
with this service disabled (1) or enabled (0). If started as disabled, the
service can be later enabled with
.I poundctl
(8).
.TP
\fBBackEnd\fR
Directives enclosed between a
.I BackEnd
and
the following
.I End
directives define a single back-end server (see below for details). You may define
multiple back-ends per service, in which case
.B Pound
will attempt to load-balance between them.
.TP
\fB[Redirect | RedirectAppend | RedirectDynamic]\fR [code] "url"
This is a special type of back-end. Instead of sending the request to a back-end
.B Pound
replies immediately with a redirection to the given URL. You may define multiple
redirectors in a service, as well as mixing them with regular back-ends.
.IP
The address the client is redirected to is determined by the command you specify.
If you specify \fBRedirect\fR, the url is taken as an absolute host and path
to redirect to.  If you use \fBRedirectAppend\fR, the original request path
will be appended to the host and path you specified.  If you use \fBRedirectDynamic\fR,
then
.I url
can contain RegEx replacements in the form
.I $1
through
.I $9
which indicate expression captured from the original request path. You must have a
\fBURL\fR directive, and the first \fBURL\fR directive for the service is the one
used for capturing expressions.
.IP
Examples: if you specified
.br

.br
    Redirect "http://abc.example"
.br

.br
and the client requested
.I http://xyz/a/b/c
then it will be redirected to
.IR "http://abc.example",
but if you specified
.br

.br
    RedirectAppend "http://abc.example"
.br

.br
it will be sent to
.IR "http://abc.example/a/b/c.
.IP
If you specified
.br
    URL "^/a(/([^/]*)(/[^/]*)"
.br
    RedirectDynamic "http://abc.example$2$1/index.html"
.br

.br
it will be sent to
.IR "http://abc.example/c/b/index.html.
.IP
.IR "Technical note":
in an ideal world
.B Pound
should reply with a "307 Temporary Redirect" status. Unfortunately, that is not
yet supported by all clients (in particular HTTP 1.0 ones), so
.B Pound
currently replies by default with a "302 Found" instead. You may override this
behaviour by specifying the code to be used (301, 302 or 307).
.TP
\fBEmergency\fR
Directives enclosed between an
.I Emergency
and
the following
.I End
directives define an emergency back-end server (see below for details). You may define
only one emergency server per service, which
.B Pound
will attempt to use if all backends are down.
.TP
\fBSession\fR
Directives enclosed between a
.I Session
and
the following
.I End
directives define a session-tracking mechanism for the current service. See below
for details.
.SH "BackEnd"
A back-end is a definition of a single back-end server
.B Pound
will use to reply to incoming requests.  All configuration directives enclosed between
.I BackEnd
and
.I End
are specific to a single service. The following directives are available:
.TP
\fBAddress\fR address
The address that
.B Pound
will connect to. This can be a numeric IP address, or a symbolic host name
that must be resolvable at run-time. If the name cannot be resolved to a valid
address,
.B Pound
will assume that it represents the path for a Unix-domain socket. This is a
.B mandatory
parameter.
.TP
\fBPort\fR port
The port number that
.B Pound
will connect to. This is a
.B mandatory
parameter for non Unix-domain back-ends.
.TP
\fBHTTPS\fR
The back-end is using HTTPS.
.TP
\fBCert\fR "certificate file"
Specify the certificate that
.B Pound
will use as a client. The
.I certificate file
is the file containing the certificate, possibly a certificate chain and the signature.
This directive may appear only after the
.I HTTPS
directive.
.TP
\fBDisable\fR SSLv2|SSLv3|TLSv1|TLSv1_1|TLSv1_2
Disable the protocol \fBand all lower protocols as well\fR.
This is due to a limitation in OpenSSL, which does not support disabling a single
protocol. For example,
.I Disable TLSv1
would disable SSLv2, SSLv3 and TLSv1, thus allowing only TLSv1_1 and TLSv1_2.
This directive may appear only after the
.I HTTPS
directive.
.TP
\fBCiphers\fR "acceptable:cipher:list"
This is the list of ciphers that will be accepted by the SSL connection; it is a
string in the same format as in OpenSSL
.I ciphers(1)
and
.I SSL_CTX_set_cipher_list(3).
This directive may appear only after the
.I HTTPS
directive.
.TP
\fBPriority\fR val
The priority of this back-end (between 1 and 9, 5 is default). Higher priority
back-ends will be used more often than lower priority ones, so you should
define higher priorities for more capable servers.
.TP
\fBTimeOut\fR val
Override the global
.I TimeOut
value.
.TP
\fBConnTO\fR val
Override the global
.I ConnTO
value.
.TP
\fBWSTimeOut\fR val
Override the global
.I WSTimeOut
value.
.TP
\fBHAport\fR [ address ] port
A port (and optional address) to be used for server function checks. See below
the "High Availability" section for a more detailed discussion. By default
.B Pound
uses the same address as the back-end server, but you may use a separate address
if you wish. This directive applies only to non Unix-domain servers.
.TP
\fBDisabled\fR 0|1
Start
.B Pound
with this back-end disabled (1) or enabled (0). If started as disabled, the
back-end can be later enabled with
.I poundctl
(8).
.SH "Emergency"
The emergency server will be used once all existing back-ends are "dead".
All configuration directives enclosed between
.I Emergency
and
.I End
are specific to a single service. The following directives are available:
.TP
\fBAddress\fR address
The address that
.B Pound
will connect to. This can be a numeric IP address, or a symbolic host name
that must be resolvable at run-time. If the name cannot be resolved to a valid
address,
.B Pound
will assume that it represents the path for a Unix-domain socket. This is a
.B mandatory
parameter.
.TP
\fBPort\fR port
The port number that
.B Pound
will connect to. This is a
.B mandatory
parameter for non Unix-domain back-ends.
.SH "Session"
Defines how a service deals with possible HTTP sessions.  All configuration
directives enclosed between
.I Session
and
.I End
are specific to a single service. Once a sessions is identified,
.B Pound
will attempt to send all requests within that session to the same back-end
server.
.PP
The following directives are available:
.TP
\fBType\fR IP|BASIC|URL|PARM|COOKIE|HEADER
What kind of sessions are we looking for: IP (the client address), BASIC (basic
authentication), URL (a request parameter), PARM (a URI parameter), COOKIE (a
certain cookie), or HEADER (a certain request header).
This is a
.B mandatory
parameter.
.TP
\fBTTL\fR seconds
How long can a session be idle (in seconds). A session that has been idle for
longer than the specified number of seconds will be discarded.
This is a
.B mandatory
parameter.
.TP
\fBID\fR "name"
The session identifier. This directive is permitted only for sessions of type
URL (the name of the request parameter we need to track), COOKIE (the name of
the cookie) and HEADER (the header name).
.PP
See below for some examples.
.SH HIGH-AVAILABILITY
.B Pound
attempts to keep track of active back-end servers, and will temporarily disable
servers that do not respond (though not necessarily dead: an overloaded server
that
.B Pound
cannot establish a connection to will be considered dead). However, every
.I Alive
seconds, an attempt is made to connect to the dead servers in case they have become
active again. If this attempt succeeds, connections will be initiated to them again.
.PP
In general it is a good idea to set this time interval as low as is consistent with
your resources in order to benefit from resurected servers at the earliest possible
time. The default value of 30 seconds is probably a good choice.
.PP
The clients that happen upon a dead backend server will just receive a
.I "503 Service Unavailable"
message.
.PP
The
.I HAport
parameter specifies an additional port (and optionally an address)
that is used only for viability checks: if this port is specified in a
.I BackEnd
directive,
.B Pound
will attempt periodically (every
.I Alive
seconds) to connect to this port. If the port does not respond the server is considered dead.
.B "It never makes sense to have the"
.I HAport
.B "identical to the main back-end port:"
this would only generate extra, unncecessary activity (CPU, network traffic) for no good
reason whatsoever.  The
.I HAport
is meant for applications that offer an additional health monitoring port or for installations
that wish to take servers off-line in a controlled manner.
.PP
By default the address of the
.I HAport
health monitor is the same as that of the
back-end server. You may specify a different address though, for example if you have
a monitoring program running on another host.
.SH HTTPS HEADERS
If a client browser connects to
.B Pound
via HTTPS and if it presents a client certificate
.B Pound
adds the following headers to the request it issues to the server:
.TP
\fBX-SSL-Subject\fR
Details about the certificate owner.
.TP
\fBX-SSL-Issuer\fR
Details about the certificate issuer (Certificate Authority).
.TP
\fBX-SSL-notBefore\fR
Starting date of certificate validity.
.TP
\fBX-SSL-notAfter\fR
Ending date of certificate validity.
.TP
\fBX-SSL-serial\fR
Certificate serial number (decimal).
.TP
\fBX-SSL-cipher\fR
The cipher currently in use.
.TP
\fBX-SSL-certificate\fR
The full client certificate (PEM-format multi-line)
.PP
It is the application's responsibility to actually use these
headers - Pound just passes this information without checking
it in any way (except for signature and encryption correctness).
.SH SECURITY
.PP
In general,
.B Pound
does not read or write to the hard-disk. The exceptions are reading the configuration file
and (possibly) the server certificate file(s) and error message(s), which are opened read-only
on startup, read,
and closed, and the pid file which is opened on start-up, written to and immediately closed.
Following this there is no disk access whatsoever, so using a RootJail directive is only
for extra security bonus points.
.PP
.B Pound
tries to sanitise all HTTP/HTTPS requests: the request itself, the headers and the contents
are checked for conformance to the RFC's and only valid requests are passed to the back-end
servers. This is not absolutely fool-proof - as the recent Apache problem with chunked
transfers demonstrated. However, given the current standards, this is the best that can
be done - HTTP is an inherently weak protocol.
.SH ADDITIONAL NOTES
.B Pound
uses the system log for messages (default facility LOG_DAEMON). The format is very similar to
other web servers, so that if you want to use a log tool:
.TP
    fgrep pound /var/log/messages | your_log_tool
.PP
Translating HTTPS to HTTP is an iffy proposition: no client information is passed to
the server itself (certificates, etc) and the backend server may be misled if it
uses absolute URLs. A patch for \fIZope\fR is included in the distribution to address
this issue - for other Web servers you are on your own. May the source be with you.
.PP
.B Pound
deals with (and sanitizes) HTTP/1.1 requests. Thus even if you have an HTTP/1.0 server,
a single connection to an HTTP/1.1 client is kept, while the connection to the back-end
server is re-opened as necessary.
.PP
.B Pound
attempts to resolve the names of the hosts that appear in various requests and/or responses.
That means it need a functioning resolver of some kind (be it /etc/hosts, DNS or something
else).
.SH EXAMPLES
To translate HTTPS requests to a local HTTP server (assuming your network address
is 123.123.123.123):
.IP
ListenHTTPS
.br
    Address 1.2.3.4
.br
    Port    443
.br
    Cert    "/etc/pound/server.pem"
.br

.br
    Service
.br
        BackEnd
.br
            Address 127.0.0.1
.br
            Port    80
.br
        End
.br
    End
.br
End
.PP
To distribute the HTTP/HTTPS requests to three Web servers, where the third one
is a newer and faster machine:
.IP
ListenHTTP
.br
    Address 123.123.123.123
.br
    Port    80
.br
End
.br
ListenHTTPS
.br
    Address 1.2.3.4
.br
    Port    443
.br
    Cert    "/etc/pound/server.pem"
.br
End
.br

.br
Service
.br
    BackEnd
.br
        Address 192.168.0.10
.br
        Port    80
.br
    End
.br
    BackEnd
.br
        Address 192.168.0.11
.br
        Port    80
.br
    End
.br
    BackEnd
.br
        Address 192.168.0.12
.br
        Port    80
.br
        Priority 3
.br
    End
.br
End
.PP
To separate between image requests and other Web content and send all requests
for a specific URL to a secure server:
.IP
ListenHTTP
.br
    Address 123.123.123.123
.br
    Port    80
.br
End
.br

.br
# Images server(s)
.br
Service
.br
    URL ".*.(jpg|gif)"
.br
    BackEnd
.br
        Address 192.168.0.12
.br
        Port    80
.br
    End
.br
End
.br

.br
# redirect all requests for /forbidden
.br
Service
.br
    Url         "/forbidden.*"
.br
    Redirect    "https://xyzzy.com"
.br
End
.br

.br
# Catch-all server(s)
.br
Service
.br
    BackEnd
.br
        Address 192.168.0.10
.br
        Port    80
.br
    End
.br
    BackEnd
.br
        Address 192.168.0.11
.br
        Port    80
.br
    End
.br
    Session
.br
        Type    BASIC
.br
        TTL     300
.br
    End
.br
End
.PP
Here is a more complex example: assume your static images (GIF/JPEG) are to be served
from a single back-end 192.168.0.10. In addition, 192.168.0.11 is to do the
hosting for www.myserver.com with URL-based sessions, and 192.168.0.20 (a 1GHz PIII)
and 192.168.0.21 (800Mhz Duron) are for all other requests (cookie-based sessions).
The logging will be done by the back-end servers.  The configuration file may look like this:
.IP
User        "nobody"
.br
Group       "nogroup"
.br
RootJail    "/var/pound/jail"
.br
Alive       60
.br
LogLevel    0
.br

.br
# Main listening ports
.br
ListenHTTP
.br
    Address 1.2.3.4
.br
    Port    80
.br
    Client  10
.br
End
.br
ListenHTTPS
.br
    Address 1.2.3.4
.br
    Port    443
.br
    Cert    "/etc/pound/pound.pem"
.br
    Client  20
.br
End
.br

.br
# Image server
.br
Service
.br
    URL ".*.(jpg|gif)"
.br
    BackEnd
.br
        Address 192.168.0.10
.br
        Port    80
.br
    End
.br
End
.br

.br
# Virtual host www.myserver.com
.br
Service
.br
    URL         ".*sessid=.*"
.br
    HeadRequire "Host:.*www.myserver.com.*"
.br
    BackEnd
.br
        Address 192.168.0.11
.br
        Port    80
.br
    End
.br
    Session
.br
        Type    URL
.br
        ID      "sessid"
.br
        TTL     120
.br
    End
.br
End
.br

.br
# Everybody else
.br
Service
.br
    BackEnd
.br
        Address 192.168.0.20
.br
        Port    80
.br
        Priority 5
.br
    End
.br
    BackEnd
.br
        Address 192.168.0.21
.br
        Port    80
.br
        Priority 4
.br
    End
.br
    Session
.br
        Type    COOKIE
.br
        ID      "userid"
.br
        TTL     180
.br
    End
.br
End
.br
.SH FILES
.TP
\fI/var/run/pound.nnn\fR
this is where
.B Pound
will attempt to record its process id.
.TP
\fI/usr/local/etc/pound.cfg\fR
the default configuration file (the location may be changed when compiling - see the
F_CONF flag in the Makefile).
.TP
\fI/usr/local/etc/pound/cert.pem\fR
the certificate file(s) for HTTPS. The location must be defined in the configuration
file - this is only a suggestion. The file must contain a PEM-encoded certificate,
optionally a certificate chain from a known Certificate Authority to your server certificate
and a PEM-encoded private key (not password protected). See
.I OpenSSL(1)
for details. This file should be well protected, lest someone gets your server
private key.
.SH AUTHOR
Written by Robert Segall, Apsis GmbH.
.SH "REPORTING BUGS"
Report bugs to <roseg@apsis.ch>.
.SH COPYRIGHT
Copyright \(co 2002-2010 Apsis GmbH.
.br
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.