[BACK]Return to mirror.xml CVS log [TXT][DIR] Up to [cvs.NetBSD.org] / htdocs / docs

File: [cvs.NetBSD.org] / htdocs / docs / mirror.xml (download)

Revision 1.11, Thu Sep 3 10:09:06 2020 UTC (13 months, 2 weeks ago) by kim
Branch: MAIN
CVS Tags: HEAD
Changes since 1.10: +3 -3 lines

Need to leave in the http: prefix here for link to man.NetBSD.org

<?xml version="1.0"?>
<!DOCTYPE webpage
  PUBLIC "-//NetBSD//DTD Website-based NetBSD Extension//EN"
         "http://www.NetBSD.org/XML/htdocs/lang/share/xml/website-netbsd.dtd">

<webpage id="docs-mirror">
  <config param="desc" value="NetBSD Mirror FAQ"/>
  <config param="cvstag"
    value="$NetBSD: mirror.xml,v 1.11 2020/09/03 10:09:06 kim Exp $"/>
  <config param="rcsdate" value="$Date: 2020/09/03 10:09:06 $"/>

  <head>

    <!-- Copyright (c) 1994-2011
    The NetBSD Foundation, Inc.  ALL RIGHTS RESERVED. -->

    <!--
    ==  created by Jun-ichiro itojun Hagino
    -->

    <title>NetBSD Mirror FAQ</title>
  </head>

  <sect1 role="toc">

    <sect2 id="ftp-mirror">
      <title>FTP and ISO Mirror FAQ</title>

      <sect3 id="ftp-overview">
	<title>Overview for setting up a NetBSD FTP/ISO Mirror</title>

	<para>The basic steps for setting up a NetBSD FTP or ISO Mirror
	  are:</para>

	<orderedlist>
	  <listitem><simpara>Prepare disk space.  See "<ulink
	      url="#ftp-diskspace">How much disk space is needed
	      for an FTP/ISO mirror</ulink>"
	    entry below for information on amount of space
	    required.</simpara></listitem>
	  <listitem><simpara>Retrieve a copy of the NetBSD FTP site
	    contents</simpara></listitem>
	  <listitem><simpara>Set things up to synchronize the contents of your
	    mirror with <code>ftp.NetBSD.org</code>.</simpara></listitem>

	  <listitem><simpara>Send in the information about your mirror to
	    <email>mirrors@NetBSD.org</email>. See "<ulink
	      url="#ftp-contactinfo">Contact information format for
	      FTP/ISO Mirrors</ulink>"
	    for details on that.</simpara></listitem>
	</orderedlist>
      </sect3>

      <sect3 id="ftp-diskspace">
	<title>How much disk space is needed for an FTP/ISO mirror?</title>

	<para>The amount of disk space needed fluctuates over time.
	  At this time (July 2018) a complete mirror of the supported
	  NetBSD releases needs approximately 250 GB (and an additional
          520 GB for unsupported historical releases). The daily builds of
          NetBSD amount to 1.3 TB and the whole of pkgsrc (source and binary
          packages for several branches) to 1.3 TB.</para>

	<para>Space requirements for ISO files are dependent on
	  how many binary releases are supported at the time. The
	  size may be severely decreased when old branches become
	  obsolete and will grow when new versions of NetBSD are
	  released. Each release takes from 20 to 30 GB, and
	  currently (July 2018) the complete iso area needs around
	  225 GB space.</para>
      </sect3>

      <sect3 id="ftp-bandwidth">
	<title>How much bandwidth does offering an FTP/ISO mirror require?</title>

	<para>At the moment, we do not have any detailed numbers on bandwidth
          usage from our mirrors.  The actual usage of any mirror depends too
          heavily on a variety of factors such as already available bandwidth,
          geographical location relative to other mirrors and amount of data
          mirrored for us to provide you with accurate numbers.</para>

        <para>If you have been running a full mirror for a while and have
          collected bandwidth statistics, please <ulink
          url="http://www.NetBSD.org/cgi-bin/feedback.cgi">let us
          know</ulink>.</para>
      </sect3>


      <sect3 id="ftp-retrieve-sync">
	<title>Retrieving/Syncing a copy of the NetBSD FTP site
	  and ISO images area</title>

	<para>There are several ways that you can retrieve and sync
	  the files. Depending on your personal preference, your
	  network connection, the packages installed on your systems
	  etc. you may choose one over the other. Either way, it's
	  probably a good idea to first run the relevant command by
	  hand and then, when you have confirmed that everything is
	  working as it should, create a script to be run by
	  &man.cron.8;.</para>

	<para>All ISO images of the NetBSD distributions are stored on
	  the FTP site inside <filename>images/</filename> subdirectory.
	  Thus, mirroring of ISO images is very similar to mirroring
	  the entire FTP site.</para>

	<para>Below you can find some examples of how to keep your
	  local copy in sync. Simply replace "&lt;FTPROOT&gt;" with
	  the actual root of your ftp server.</para>

	<itemizedlist>
	  <listitem><para>Using <filename role="pkg">net/rsync</filename>:
	    <screen>$ rsync -v -Wartz --delete rsync://rsync.NetBSD.org/NetBSD &lt;FTPROOT&gt;/pub/NetBSD/</screen>
	    For ISO files only:
	    <screen>$ rsync -v -Wartz --delete rsync://rsync.NetBSD.org/NetBSD/images &lt;FTPROOT&gt;/pub/NetBSD/images</screen></para></listitem>
	  <listitem><para>Using <filename role="pkg">net/wget</filename>:
	    <screen>$ cd &lt;FTPROOT&gt;
$ wget -v -xnH -m ftp://ftp.NetBSD.org/pub/NetBSD/</screen>
	    For ISO files only:
	    <screen>$ cd &lt;FTPROOT&gt;
$ wget -v -xnH -m ftp://ftp.NetBSD.org/pub/NetBSD/images/</screen></para></listitem>
	  <listitem><para>Using <filename role="pkg">net/mirror</filename>:
	    <screen>$ cat &gt; ftp.NetBSD.org &lt;&lt;EOF
package=netbsd
comment=NetBSD
site=ftp.NetBSD.org
remote_dir=/pub/NetBSD
local_dir=&lt;FTPROOT&gt;/pub/NetBSD
do_deletes=true
algorithm=1
EOF
$ mirror ftp.NetBSD.org</screen>
	    For ISO files only:
	    <screen>$ cat &gt; iso.NetBSD.org &lt;&lt;EOF
package=netbsd
comment=NetBSD
site=iso.NetBSD.org
remote_dir=/pub/NetBSD/mages
local_dir=&lt;FTPROOT&gt;/pub/NetBSD/images
do_deletes=true
algorithm=1
EOF
$ mirror iso.NetBSD.org</screen></para></listitem>
	</itemizedlist>
      </sect3>

      <sect3 id="ftp-notes">
	<title>Notes about a NetBSD FTP/ISO mirror</title>

	<para>As a NetBSD FTP/ISO mirror, it's important that you
	  synchronize your copy of the NetBSD FTP tree and ISO
	  images area regularly. Otherwise, old and/or obsolete
	  files may confuse users.</para>

	<para>Also, it might happen that due to, e.g., a security
	  issue, file removal (or another administrative task) may be
	  requested from the NetBSD Project.</para>
      </sect3>

      <sect3 id="ftp-outage">
	<title>Where should an outage be announced</title>

	<para>If your FTP/ISO mirror experiences any significant system
	  outage, you should notify <email>mirrors@NetBSD.org</email>
	  as soon as possible so that we can temporarily redirect your
	  CNAME (if there is one) to a nearby alternative. Also, you
	  may wish to send a note about the outage to the <ulink
	    url="../mailinglists/#netbsd-users">netbsd-users mailing list</ulink>.</para>

	<para>If the CNAME for your mirror has been redirected, be
	  sure to let us know when the situation is rectified, so we
	  can revert the change.</para>
      </sect3>

      <sect3 id="ftp-contactinfo">
	<title>Contact information format for FTP/ISO mirrors</title>

	<para>Contact information needs to be submitted to
	  <email>mirrors@NetBSD.org</email>.</para>

	<para>This information gets added to an internal mirrors
	  database, from which the mirror-related documentation gets
	  generated. The "<code>contact*</code>" fields are never made
	  public. Contact information for NetBSD FTP/ISO mirrors should be
	  sent in the following format:</para>

	<screen>mirror          *machine name*
country         *ISO 3166 country code of host, see /usr/share/zoneinfo/iso3166.tab*
location        *more specific location (e.g, ``University of Foo'')*
ftp		*url to NetBSD ftp mirror area*
ftp-host	*fqdn of the host providing ftp mirror service*
ftp-releng	*url to NetBSD release engineering files, aka NetBSD-daily*
iso		*url to NetBSD iso images area*
iso-host	*fqdn of the host providing iso images mirror service*
organisation    *organisation name*
contactperson   *contact person's name*
contactemail    *contact person's electronic mail address*
contactkeyid	*contact person's pgp key id*
contactphone    *contact person's telephone number*
contactaddress  *contact person's postal address*</screen>

	<para>An example submission might look like this.</para>
	<screen>mirror          netbsd.foo.com
country         jp
location        Department of Computer Science, Foo University
ftp             ftp://ftp.foo-univ.ac.jp/pub/NetBSD/
ftp-host        ftp.foo-univ.ac.jp
iso		ftp://iso.foo-univ.ac.jp/pub/NetBSD/images/
iso-host	iso.foo-univ.ac.jp
organisation    Foo University
contactperson   Foo Bar
contactemail    Bar@foo-univ.ac.jp
contactkeyid	GA23FG12
contactphone    +81 3 xxxx yyyy
contactaddress  9876 Foo City, Tokyo 1234567 Japan</screen>

	<para>Please use <literal>ftp*</literal> and
	  <literal>iso*</literal> fileds in case only if you
	  providing appropriate mirror services.  For example, omit
	  <literal>ftp-releng</literal> field if there are no releng
	  files on your server, or don't use any of <literal>ftp*</literal>
	  fields if you running ISO mirror only.</para>

	<para>If the contact information for your mirror ever needs to
	  be changed, please feel free to let us know at
	  <email>mirrors@NetBSD.org</email>.</para>
      </sect3>

      <sect3 id="ftp-cname">
	<title>How to get (ftp|iso).country-code.NetBSD.org</title>

	<para>We have temporarily stopped assigning names in the
	  .NetBSD.org domain to mirrors as we are re-evaluating our
	  mirror policies.</para>
      </sect3>

      <sect3 id="ftp-teardown">
	<title>If you can no longer provide a FTP/ISO mirror</title>

	<para>Please let us know at <email>mirrors@NetBSD.org</email>
	  so that we can remove your site from our list and redirect
	  the DNS name to an alternative site.</para>
      </sect3>
    </sect2>

    <sect2 id="bitorrent">
      <title>BitTorrent Mirror FAQ</title>

      <sect3 id="bittorrent-what">
	<title>Bit-what?</title>

	<para><ulink
	    url="http://bitconjurer.org/BitTorrent/">BitTorrent</ulink>
	  is a tool for distributing files. It's extremely easy to
	  use - downloads are started by clicking on hyperlinks.
	  Whenever more than one person is downloading at once
	  they send pieces of the file(s) to each other, thus
	  relieving the central server's bandwidth burden.  Even with
	  many simultaneous downloads, the upload burden on the
	  central server remains quite small, since each new
	  downloader introduces new upload capacity.</para>
      </sect3>

      <sect3 id="bittorrent-overview">
	<title>Overview for setting up a NetBSD Bittorrent Mirror</title>

	<para>The basic steps for setting up a NetBSD BitTorrent
	  Mirror are:</para>

	<orderedlist>
	  <listitem><simpara>Install <filename
	      role="pkg">net/bittorrent</filename></simpara></listitem>
	  <listitem><simpara>Fetch the <filename>.torrent</filename> files from
	    our FTP site. The files are in the
	    <filename>/pub/NetBSD/images/&lt;release&gt;</filename>
	    subdirectory, so if you already are mirroring this
	    directory, you're all set. For more information see <ulink
	      url="../mirrors/torrents/">this page</ulink>.</simpara></listitem>
	  <listitem><para>Start BitTorrent:
	    <screen>cd /ftp/pub/NetBSD/images/&lt;release&gt;; (btlaunchmany.py .) &gt;/dev/null 2&gt;/dev/null &amp;</screen>
	    If you didn't download the ISO images already, they will be
	    downloaded now. You can use the
	    <code>--max_upload_rate arg</code>
	    and <code>--max_download_rate arg</code>
	    options to limit the transfer rates.</para></listitem>
	</orderedlist>
      </sect3>

      <sect3 id="bittorrent-contactinfo">
	<title>Contact information format for BitTorrent mirrors</title>

	<para>Contact information needs to be submitted to
	  <email>mirrors@NetBSD.org</email>.
	  This information gets added to an internal mirrors database,
	  from which the mirror-related documentation gets generated.
	  The "<code>contact*</code>" fields are never made public.
	  Contact information for NetBSD BitTorrent mirrors should be
	  sent in the following format:</para>

	<screen>mirror          *machine name*
country         *ISO 3166 country code of host, see /usr/share/zoneinfo/iso3166.tab*
location        *more specific location (e.g, ``University of Foo'')*
bittorrent	*BitTorrent URL / URL with additional Information*
ipv6		*yes or no, if your server supports IPv6*
organisation    *organisation name*
contactperson   *contact person's name*
contactemail    *contact person's electronic mail address*
contactkeyid	*contact person's pgp key id*
contactphone    *contact person's telephone number*
contactaddress  *contact person's postal address*</screen>

	<para>An example submission might look like this.</para>

	<screen>mirror          ftp.NetBSD.org
country         us
location        Silicon Valley, California
bittorrent	http://www.NetBSD.org/mirrors/torrents/
ipv6            yes
organisation    The NetBSD Foundation, Inc.
contactperson   Admins
contactemail    &lt;email-here&gt;@NetBSD.org
contactkeyid	AF226A4C
contactphone    555-123-1234
contactaddress 	Somwhere in Silicon Valley, California</screen>
      </sect3>
    </sect2>

    <sect2 id="www">
      <title>WWW Mirror FAQ</title>

      <sect3 id="www-overview">
	<title>Overview for setting up a NetBSD WWW Mirror</title>

	<para>The following is an overview of the recommended
	  procedure for setting up a NetBSD WWW mirror:</para>

	<orderedlist>
	  <listitem><simpara>For setting up a mirror with this procedure,
	    you'll need cvs, ssh, and wget
            (see <filename role="pkg">net/wget</filename>).  Of course you'll
            also need a web server, such as apache.</simpara></listitem>
	  <listitem><simpara>Retrieve a copy of the NetBSD WWW site.  See "<ulink
	      url="#www-retrieve">Retrieving a copy of the NetBSD WWW site</ulink>"
	    for instructions on how to do so with anoncvs.</simpara></listitem>
	  <listitem><simpara>Set things up to synchronize the contents of your
	    mirror with <code>www.NetBSD.org</code>.
	    See "<ulink url="#www-sync">Syncing the NetBSD WWW site contents</ulink>"
	    for an example of how to do so using a cron job.</simpara></listitem>
	  <listitem><simpara>Send in the information about your mirror to
	    <email>mirrors@NetBSD.org</email>. See
	    "<ulink url="#www-contactinfo">Contact information format for WWW Mirrors</ulink>"
	    for details on that.</simpara></listitem>
	</orderedlist>
      </sect3>

      <sect3 id="www-diskspace">
	<title>How much disk space is needed for a WWW mirror?</title>

	<para>The amount of diskspace needed for a complete copy of
	  the htdocs repository is currently (July 2018) around 310 MB.
	  The total amount should not vary too much, but will
	  obviously grow with time.</para>
      </sect3>

      <sect3 id="www-retrieve">
	<title>Retrieving a copy of the NetBSD WWW site</title>

	<para>To retrieve a copy of <code>www.NetBSD.org</code> via
	  anoncvs, you should do:
	  <screen># su - nbwww
# cd /path/to/htdocs
# env CVS_RSH=ssh CVSROOT=anoncvs@anoncvs.NetBSD.org:/cvsroot \
  cvs checkout -P htdocs</screen>

	  <emphasis role="bold">Note:</emphasis>
	  "<filename>/path/to/htdocs</filename>"
	  should be where you want the mirror of
	  <code>www.NetBSD.org</code> to reside.  The directory
	  should be owned by an e.g. "nbwww" user.</para>
      </sect3>

      <sect3 id="www-sync">
	<title>Syncing the NetBSD WWW site contents</title>

	<para>The best way to do this is to set up a cron job for
	  your e.g. nbwww user to update the web site at a regular
	  interval using anoncvs and wget (more on the wget usage
	  below):
	  <screen id="www-mirror-commands"># crontab -u nbwww -e
SHELL=/bin/sh
PATH=/bin:/sbin:/usr/bin:/usr/sbin:/usr/pkg/bin
MAILTO=whoever@gets.the.mirror.mail
CVS_RSH=ssh
CVSROOT=anoncvs@anoncvs.NetBSD.org:/cvsroot
0 1 * * * cd /path/to/htdocs &amp;&amp; cvs update -dP &amp;&amp; wget -B http://www.NetBSD.org/ -nv -xnH -N -Fi mirrors/fetch.html</screen>
	  <emphasis role="bold">Note:</emphasis> In the suggested
	  contents, you'll want to change "MAILTO" to reflect who
	  should get the mirror-related mail.
	  Also, "<filename>/path/to/htdocs</filename>" should be
	  changed to where your <code>www.NetBSD.org</code> mirror
	  resides.</para>
      </sect3>

      <sect3 id="www-fetch">
	<title>What is the purpose of <filename>htdocs/mirrors/fetch.html</filename></title>

	<para>The <filename>fetch.html</filename> file links to the
	  handful of documents on <code>www.NetBSD.org</code> that
	  are generated <emphasis>by</emphasis>
	  <code>www.NetBSD.org</code>.  These files get generated
	  daily from content in the <code>doc</code> module of the
	  NetBSD CVS repository.</para>

	<para>In the example cron tab above, you'll see
	  "<code>wget -nv -xnH -NFi mirrors/fetch.html</code>".
	  This means that daily, through wget, these
	  <code>www.NetBSD.org</code> created files
	  will be updated in your copy of the web site
	  contents.</para>
      </sect3>

      <sect3 id="www-http-1.0">
	<title>Ensuring HTTP/1.0 clients get the right page</title>

	<para>It is much preferred that NetBSD WWW mirror sites run
	  a separate virtual web server for a
	  <code>www.NetBSD.org</code> mirror on a separate IP
	  address.  This is so that clients using HTTP/1.0 still
	  get to the correct web site.</para>

	<para>If that's not possible in your set up, then the other
	  solution for HTTP/1.0 clients would be putting the NetBSD
	  mirror as the primary name-based virtual host.</para>
      </sect3>

      <sect3 id="www-contactinfo">
	<title>Contact information format for WWW mirrors</title>

	<para>Contact information needs to be submitted to
	  <email>mirrors@NetBSD.org</email>.
	  This information gets added to an internal mirrors
	  database, from which the mirror-related documentation gets
	  generated.  The "<code>contact*</code>" fields are never
	  made public.  Contact information for NetBSD WWW mirrors
	  should be sent in the following format:</para>
	<screen>mirror          *machine name*
country         *ISO 3166 country code of host, see /usr/share/zoneinfo/iso3166.tab*
location        *more specific location (e.g, ``University of Foo'')*
www             *WWW URL*
www-host        *external WWW name*
organisation    *organisation name*
contactperson   *contact person's name*
contactemail    *contact person's electronic mail address*
contactkeyid	*contact person's pgp key id*
contactphone    *contact person's telephone number*
contactaddress  *contact person's postal address*</screen>

	<para>An example submission might look like this.</para>

	<screen>mirror          netbsd.foo-univ.ac.jp
country         jp
location        Department of Computer Science, Foo University
www             http://netbsd.foo-univ.ac.jp/
www-host        netbsd.foo-univ.ac.jp
organisation    Foo University
contactperson   Foo Bar
contactemail    Bar@foo-univ.ac.jp
contactkeyid	GA23FG12
contactphone    +81 3 xxxx yyyy
contactaddress  9876 Foo City, Tokyo 1234567 Japan</screen>

	<para>If the contact information for your mirror ever needs
	  to be changed, please feel free to let us know at
	  <email>mirrors@NetBSD.org</email>.</para>
      </sect3>

      <sect3 id="www-cname">
	<title>How to get www.country-code.NetBSD.org</title>

	<para>We have temporarily stopped assigning names in the
	  .NetBSD.org domain to mirrors as we are re-evaluating our
	  mirror policies.</para>
      </sect3>

      <sect3 id="www-teardown">
	<title>If you can no longer provide a WWW mirror</title>

	  <para>Please let us know at
	  <email>mirrors@NetBSD.org</email> so that we can remove your
	  site from our list and redirect the DNS name to an
	  alternative site.</para>
      </sect3>
    </sect2>

    <sect2 id="cvsweb">
      <title>CVSweb Mirror FAQ</title>

      <sect3 id="cvsweb-overview">
	<title>Overview for setting up a NetBSD CVSweb Mirror</title>

	<para>The following is an overview of the recommended
	  procedure for setting up a NetBSD CVSweb mirror:</para>

	<orderedlist>
	  <listitem><simpara>For setting up a mirror with this procedure,
	    you'll need cvs, cvsup, and the cvsweb CGI script.  Both cvsup
            and the cvsweb CGI script are available from the <ulink
	      url="software/packages.html">NetBSD packages collection</ulink>
	    (see <filename role="pkg">devel/cvsup-bin</filename> and <filename
	      role="pkg">www/cvsweb</filename>).  Of course you'll also
	    need a web server, such as <filename
	      role="pkg">www/apache</filename>.</simpara></listitem>
	  <listitem><simpara>Retrieve a copy of the NetBSD CVS tree.  See
	    "<ulink url="#cvsweb-retrieve">Retrieving a copy of the NetBSD CVS tree</ulink>"
	    for instructions on how to do so with cvsup.</simpara></listitem>
	  <listitem><simpara>Set things up to synchronize the contents of your
	    mirror with the cvsup server. See "<ulink
	      url="#cvsweb-sync">Syncing the NetBSD CVS tree contents</ulink>"
	    for an example of how to do so using a cron job.</simpara></listitem>
	  <listitem><simpara>Send in the information about your mirror to
	    <email>mirrors@NetBSD.org</email>. See  "<ulink
	      url="#cvsweb-contactinfo">Contact information format for CVSweb Mirrors</ulink>"
	    for details on that.</simpara></listitem>
	  </orderedlist>
      </sect3>

      <sect3 id="cvsweb-diskspace">
	<title>How much disk space is needed for a CVSweb mirror?</title>

	<para>The amount of disk space needed is rather large due to
	  the fact that you are providing an interface to the
	  <emphasis>entire CVS repository</emphasis>. Currently
	  (2006-05), the total disk space required for all sources in
	  CVS is around 4.3 GB.</para>
      </sect3>

      <sect3 id="cvsweb-retrieve">
	<title>Retrieving a copy of the NetBSD CVS tree</title>

	<para>To retrieve a copy of the CVS tree via cvsup, you
	  should do:
	  <screen># su - cvsweb
# mkdir -p /home/cvsweb/netbsd
# cd /home/cvsweb</screen>

	  <emphasis role="bold">Note:</emphasis>
	  "<filename>/home/cvsweb/netbsd</filename>"
	  should be where you want the mirror of the tree to reside.
	  The directory should be owned by an e.g. "cvsweb"
	  user. Create an "supfile" using your editor (For examples,
	  see <ulink url="netbsd-supfile">netbsd-supfile</ulink>).
	  Please take a look at the list of available public cvsup
	  servers, you'll find the list <ulink
	    url="../mirrors/index.html#cvsup">here</ulink>.
	  <screen># cvsup -g -L 2 netbsd-supfile</screen>
	</para>
      </sect3>

      <sect3 id="cvsweb-sync">
	<title>Syncing the NetBSD CVS tree contents</title>

	<para>The best way to do this is to set up a cron job for your
	  e.g. cvsweb user to update the tree at a regular interval
	  using cvsup:
	  <screen># crontab -u cvsweb -e
SHELL=/bin/sh
PATH=/bin:/sbin:/usr/bin:/usr/sbin:/usr/pkg/bin
MAILTO=whoever@gets.the.mirror.mail
0 1 * * * cvsup -g /home/cvsweb/netbsd-supfile</screen>

	  <emphasis role="bold">Note:</emphasis> In the suggested
	  contents, you'll want to change "MAILTO" to reflect who
	  should get the mirror-related mail.</para>
      </sect3>

      <sect3 id="cvsweb-cgi">
	<title>How to setup the cvsweb CGI script</title>

	<para>Take a look at the <filename>cvsweb.conf</filename> and
	  be sure to edit the following options:</para>
	  <screen>
@CVSrepositories = (
        'netbsd'  => ['NetBSD',           '/home/cvsweb/netbsd'],
);

$mancgi = "http://man.NetBSD.org/%s.%s";
</screen>
	  <para>Place cvsweb.cgi in your web servers' cgi-bin
	  directory and be sure to edit it to point to your
	  <filename>cvsweb.conf</filename>.</para>
      </sect3>

      <sect3 id="cvsweb-http-1.0">
	<title>Ensuring HTTP/1.0 clients get the right page</title>

	<para>It is much preferred that NetBSD CVSweb mirror sites run
	  a separate virtual web server for a CVSweb mirror on a
	  separate IP address.  This is so that clients using HTTP/1.0
	  still get to the correct web site.</para>

	<para>If that's not possible in your set up, then the other
	  solution for HTTP/1.0 clients would be putting the NetBSD
	  mirror as the primary name-based virtual host.</para>
      </sect3>

      <sect3 id="cvsweb-contactinfo">
	<title>Contact information format for CVSweb mirrors</title>

	<para>Contact information needs to be submitted to
	  <email>mirrors@NetBSD.org</email>.

	  This information gets added to an internal mirrors database, from
	  which the mirror-related documentation gets generated. The
	  "<code>contact*</code>" fields are never made public.
	  Contact information for NetBSD CVSweb mirrors should be sent
	  in the following format:</para>

	<screen>mirror          *machine name*
country         *ISO 3166 country code of host, see /usr/share/zoneinfo/iso3166.tab*
location        *more specific location (e.g, ``University of Foo'')*
cvsweb          *CVSweb URL*
cvsweb-host     *external CVSweb name*
organisation    *organisation name*
contactperson   *contact person's name*
contactemail    *contact person's electronic mail address*
contactkeyid	*contact person's pgp key id*
contactphone    *contact person's telephone number*
contactaddress  *contact person's postal address*</screen>

	<para>An example submission might look like this.</para>

	<screen>mirror          netbsd.foo-univ.ac.jp
country         jp
location        Department of Computer Science, Foo University
cvsweb          http://netbsd.foo-univ.ac.jp/cgi-bin/cvsweb.cgi/
cvsweb-host     cvsweb.netbsd.foo-univ.ac.jp
organisation    Foo University
contactperson   Foo Bar
contactemail    Bar@foo-univ.ac.jp
contactkeyid	GA23FG12
contactphone    +81 3 xxxx yyyy
contactaddress  9876 Foo City, Tokyo 1234567 Japan</screen>

	<para>If the contact information for your mirror ever needs to
	  be changed, please feel free to let us know at
	  <email>mirrors@NetBSD.org</email>.</para>
      </sect3>

      <sect3 id="cvsweb-cname">
	<title>How to get cvsweb.country-code.NetBSD.org</title>

	<para>We have temporarily stopped assigning names in the
	  .NetBSD.org domain to mirrors as we are re-evaluating our
	  mirror policies.</para>
      </sect3>

      <sect3 id="cvsweb-teardown">
	<title>If you can no longer provide a CVSweb mirror</title>

	<para>Please let us know at <email>mirrors@NetBSD.org</email>
	  so that we can remove your site from our list and redirect the
	  DNS name to an alternative site.</para>
      </sect3>
    </sect2>
  </sect1>
</webpage>