archie/release/base/manpages/archie_protocol.5

198 lines
6.1 KiB
Groff
Raw Normal View History

2024-05-28 17:59:32 +02:00
.\" Copyright (c) 1992,1994,1996 Bunyip Information Systems Inc.
.\" All rights reserved.
.\"
.\" Archie 3.5
.\" August 1996
.\"
.\" @(#)archie_protocol.5
.\"
.TH ARCHIE_PROTOCOL 5 "August 1996"
.SH SYNOPSIS
.B Archie_protocol
\- description of the internal client/server Archie protocol
.SH DESCRIPTION
.PP
This protocol describes the \fBinternal\fP Archie protocol, not that used
between the Prospero servers and clients for querying the database.
This protocol is used for distributing data between Archie systems
worldwide.
Protocol commands themselves are in ASCII printable characters and <cr>,
<lf> (ASCII 13 and 10 respectively) although data such as that from the
Archie files database are transmitted as a binary stream in Sun XDR
format. All commands must start at the beginning of the line and are
terminated by <cr><lf>. Intervening whitespace may be spaces or tabs or
both. The data stream may be compressed depending on the configuration of
the servers at either end of the connection.
The basic security mechanism enforced by the server is such that it only
allows those clients attempting connection from "approved" hosts to
establish a session. This is described in
.BR arserver (n).
It is assumed in the following description that such a session (the
\fIcontrol connection\fP) has been
established.
The following commands are used by the client and server to communicate:
.PP
.B LISTSITES
Sent by the client to the server to request all sites matching the
criteria set out by the parameters to the command. These parameters in
order are:
.RS
<db> '<'|'>' <from date> <domainlist>
.RE
All sites which have been updated in the databases named <db> more/less
recently than <from date> are to be listed. <db> is composed of a colon
separated list of database names. The character '<' means less recently,
'>' means more recently. <from date> is a date string in the format
YYYYMMDDHHMMSS in UTC. All zeros in this field is taken to mean that all
sites in <db> are to be listed. Sites listed must also match the
<domainlist> field. This can either be '*' for any domain or a colon
separated list of real domains or pseudo domains. Pseudo-domains are
user created and reside after installation in the file
\fB~Archie/host_db/domain-db\fP. See
.BR ardomains (n)
for further explanation.
The first line of response from the server is:
.RS
TUPLELIST <number>
.RE
If <number> is non-zero the rest of the response is composed of a set
tuples, <site tuple>, one for each site/database pair in the files
matching the given criteria, one per line terminated by <cr><lf>. If
<number> is zero, then no further lines are transmitted. Each tuple
uniquely identifies that entry across the entire Archie system. The
tuples consist of the following fields in the given order:
<source>:<date>:<primary name>:<pref name>:<ip addr>:<db>
Where:
.RS
.TP
.B <source>
is the Archie host responsible for monitoring that site
.TP
.B <date>
is the date of listing retrieve of anonymous ftp host
.TP
.B <primary name>
is the primary host name of site in the database
.TP
.B <pref name>
is the preferred host name (CNAME) of host. Field may be empty if no such
name is stored
.TP
.B <ip addr>
is the primary IP address of the host
.TP
.B <db>
is the database name
.RE
Host names are specified in standard RFC 1037 format
IP addresses are specified in standard 'dotted decimal' notation
The date is specified YYYYMMDDHHMMSS in UTC. However this value is not
examined other than for inequality tests with internal records.
.PP
.B SENDSITE
With this command, the client informs the server that it would like the
information about a particular site/database combination. It determines
this by comparing the tuples returned by the LISTSITES command above with
its local database. The parameters to the command in order are:
.RS
<primary hostname>:<database>[:<port>] ["compress"]
.RE
If the "compress" string is included with the protocol command, then the
remote server is requested to compress the data stream. The remote server
may agree but also has the option of ignoring this request.
In the case of the `webindex' database, the port is specified, as multiple
servers can reside on the `<primary hostname>' machine.
The server responds with:
.RS
SITELIST <port number>
.RE
With this response the server informs the client that it ready to
transmit the information and that it is available on <port number>. The
client then opens a connection (data channel) on the server host with
that <port number>.
The format in which the information for this site/database combination is
transmitted is determined by the actual database information. See
.BR arserver (n).
The header for that site/database is usually the first piece of
information to be transmitted on the data channel, however this is not
defined by the protocol.
There is no acknowledgement from the client.
.PP
.B SENDHEADER
With this command the client specifies to the server that the header
record should be transmitted on the control connection. This command is
used by the client when invoked in "retrieval mode" (see
.BR arserver (n).
The parameters with this command are (in order):
.RS
<primary hostname>:<database>
.RE
There is no acknowledgement from the client.
.PP
.B SENDEXCERPT
With this command the client specifies to the server that the excerpt record
should be transmitted on the data connection. This command is used by the
client after receiving the site for the `webindex' database.
This command creates the sam reply from the server as if it was a
SENDSITE command.
The parameters with this command are (in order):
.RS
<primary hostname>:<database>:<port>
.RE
.PP
.B DUMPCONFIG
This command causes the server to list its arupdate.cf configuration
file. The format is the same as that used in the configuration file
except that semicolons (':') are used as field separators. The server
signals the termination of the output by the line ENDDUMP.
.PP
.B QUIT
The client requests that the control connection be closed. There is no
acknowledgement from the server.
.SH "SEE ALSO"
.BR arserver (n)
.SH AUTHOR
Bunyip Information Systems
.br
Montr\o"\'e"al, Qu\o"\'e"bec, Canada
.sp
Archie is a registered trademark of Bunyip Information Systems Inc., Canada,
1990.