table of contents
RLINETD.CONF(5) | rlinetd 0.9 | RLINETD.CONF(5) |
NAME¶
rlinetd.conf - rlinetd configuration fileDESCRIPTION¶
rlinetd.conf holds configuration information for rlinetd. There are a small number of similar top level constructs, differing chiefly in which options can be meaningfully used with them. All strings are quoted with the " character. In some situations (e.g. the log, exec, and chroot directives), there are a number of variables that can be substituted into the string....
}
This construct describes a service. The name
parameter is for naming convenience alone, it simply serves to distinguish
logging messages and provides a default for options which can logically accept
a name as an argument.
enabled
log "name" {
This construct allows easily enabling or disabling
service. The argument can be either yes or no. The default value
is yes. Setting this to no disables service.
Example:
port
enabled no;
This lists the ports that the service should be made
available on. The ports can be listed in either string or numeric format. If
unspecified, this defaults to the name of the service unless the service is an
RPC service, in which case the port value will be dynamically assigned by the
system.
Example:
interface
port "telnet", "rcmd", 56,
99;
This specifies which interfaces the listed ports should
be bound on. It takes a list of IP addresses as an argument, corresponding to
the configured addresses of the interfaces required. If this keyword is not
given or when a special value any is assigned to it, the service will
bind to all available interfaces.
Examples:
exec
interface 192.168.1.1, 192.168.1.2;
interface any;
This specifies the invocation of the service. A number of
substitutions can be made within the string; please see String
Modifiers below.
Example:
server
exec "/usr/sbin/in.telnetd -d";
This specifies the binary to be executed, if different to
exec.
Example:
protocol
server "/usr/sbin/tcpd";
This specifies the socket protocol to use when listening
on ports for the service. The argument can be either tcp or udp.
The default setting for this variable is tcp.
Example:
user
protocol tcp;
This specifies the userid under which this service should
run. It will accept an argument in either symbolic or numeric form. Unless
group (see below) is given, the groupid is also set to the user's
primary group.
Example:
group
user "nobody";
This specifies the groupid under which this service
should run. It will accept an argument in either symbolic or numeric form.
Example:
backlog
group "system";
This is the backlog argument which will be passed to the
listen(2) system call.
Example:
instances
backlog 30;
This specifies the maximum number of service instances
that can be running at any one time. The default setting for this variable is
40.
Example:
wait
instances 50;
This directive emulates the inetd(8) wait
behaviour. The argument can be either yes or no. The default
value is no. Setting this to yes also resets the value of
instances option to 1.
Example:
nice
wait yes;
This specifies the process priority to run this service
at. The argument is passed directly to the setpriority(2) system call.
The value may be negative.
Example:
rpc
nice -5;
This specifies that the service should be registered with
the system's portmap(8) mapper as an RPC service. It accepts a list of
arguments as follows.
rpc {
chroot
name "string"; version
3,6,9-15,22;
}
The name parameter is optional, and defaults to the service name.This specifies the root directory for the service. The
string argument can accept modifiers as detailed in String Modifiers
below.
Example:
log
chroot "/tftpboot/%O";
This directive takes two arguments. The first must be
either the symbolic name of a previously specified log directive (see
below), or the unquoted word syslog. If the latter, the message will be
logged via the syslog(3) call. The second argument is the message that
will be logged, subject to the modifiers detailed in String Modifiers
below.
Example:
tcpd
log syslog "Service from %O
complete";
This directive causes access controls as specified by
tcp_wrappers to be applied. This has the same effect as invoking a service
with a server argument of /usr/sbin/tcpd (or wherever your
tcpd(8) program is kept), but saves the additional step of starting the
program. It will accept up to two additional arguments. The first is a service
name to apply against its rules, and the second is a block of instructions to
execute if matched. If no name is specified, it defaults to the name of the
service. If the instruction block is not specified, it defaults to 'exit;'.
Examples:
exit
tcpd "in.telnetd";
tcpd { exec "/usr/local/bin/winnuke %O"; }
tcpd "pointless" { echo "Hi guys, come on in."; }
tcpd "defiant" { echo "500 Access denied from %O."; exit; }
This directive is only useful in an instruction block
argument to the tcpd directive. Note well - not using it (and not
specifying some other terminating directive, such as exec) will result
in the service being run forever.
Example:
capability
exit;
This directive specifies the capabilities that this
service should have when running. The argument is a string that is passed
directly to cap_from_text(3). I know, that's a pretty lousy
description, but this feature is of limited utility until and unless you read
the README.capabilities file anyway.
Example:
rlimit
capability "cap_setuid=ep";
This directive takes two arguments. The first is a symbol
specifying the type of limit required. These are listed below. The second
argument takes one of two forms. It can either be a single numeric value, in
which case both of the soft and hard limits of the resource in question will
be set to this value. Alternatively, it can be a list in the form:
rlimit type {
initgroups
soft x; hard y;
}
In which case the hard and soft limits will be set appropriately. In either
case, the word unlimited can be specified instead of a numeric value,
thus removing any restriction. The values are passed directly to the
setrlimit(2) syscall, and should be specified in that context.
Types:
cpu, fsize, data, stack, core, rss, nproc, nofile,
memlock
Example:
rlimit cpu 15;
The argument can be either yes or no. This
directive causes initgroups(3) to be called at service startup, which
sets the supplementary groups of the service according to the
/etc/group file.
Example:
family
initgroups yes;
This directive specifies the protocol family that rlinetd
should bind sockets on for this service. Currently, this can be either
ipv4 or ipv6. If unspecified, this defaults to something
appropriate for the system.
Example:
banner
family ipv6;
This directive lets you dump a file as output to a
connection.
Example:
echo
banner "/etc/nologin";
This directive allows you to output a dynamically
generated line to the connection.
Example:
filter
echo "500 Service denied from your IP
(%O)";
This directive allows you to specify a Linux Socket
Filter program to be associated with the listening socket. These can be
generated with a tool such as lsfcc(1).
Example:
chargen
filter
"/usr/local/lib/rlinetd/filters/privport";
This directive loops eternally, outputting data to any
connection. If no argument is given, it echoes a subset of the printable
characters. However, a filename can be supplied as an argument, in which case
the contents of that file are output in a loop.
Example:
chargen "/usr/local/lib/spam";
...
}
This construct describes a logging target. The
name parameter is used as an argument to the log directive in
service configurations.
path
defaults {
This specifies the filename for this logfile.
Example:
mode
path "/var/log/service.log";
This specifies the file permissions for the logfile. The
argument is required to be numeric, and defaults to 0640 if not specified.
Example:
user
mode 0600;
This specifies the uid of the logfile, and can be
specified as either a numeric uid, or username.
Example:
group
user "adm";
This specifies the gid of the logfile, and can be
specified as either a numeric gid, or groupname.
Example:
group "adm";
...
}
This construct takes the same parameters as a
service declaration, but instead of specifying a service sets defaults
for all services specified subsequently.
directory "path" "match" "ignore";
This construct specifies a directory which contains
additional configuration files to be parsed. Parsing of these additional files
does not commence until the current file is complete. The match and
ignore arguments are optional, and if specified, are used to filter the
files in the directory. Filenames must match the match regexp, if
given, and must not match the ignore regexp, if given. Filenames
beginning with a period ('.') are skipped in all cases. Directories are not
recursed into.
String Modifiers¶
There are a number of variables which can be substituted into arguments to some directives. Although they can all be used in the same places, the information accessed by some is unavailable in certain cases.- %O
- The source IP address of the connection.
- %P
- The source port of the connection.
- %C
- The total CPU time used.
- %U
- The user CPU time used.
- %S
- The system CPU time.
- %r
- Maximum resident set size.
- %m
- Shared memory size.
- %d
- Unshared data size.
- %s
- Unshared stack size.
- %f
- Page reclaims.
- %F
- Page faults.
- %p
- Swaps.
- %i
- Block input operations.
- %o
- Block output operations.
- %n
- Messages sent.
- %c
- Messages received.
- %k
- Signals received.
- %w
- Voluntary context switches.
- %w
- Involuntary context switches.
- %e
- Exit code.
- %t
- Running time.
- %M
- The current time expressed as seconds since epoch 1980, dumped as a network order 32 bit word. This has absolutely no use other than in implementing the inetd-like time functionality.
- %I
- The current time and date, in pretty-printed ctime(3) format.
SEE ALSO¶
rlinetd(8), hosts_access(5)AUTHOR¶
This manual page was written by Mikolaj J. Habryn <dichro-doc@rcpt.to>. Modified by Robert Luberda <robert@debian.org>.November 11th, 2013 | Debian |