Difference between revisions of "Mirror"

From CSCWiki
Jump to navigation Jump to search
m (add to Services and Software categories)
Line 76: Line 76:
| OpenSUSE (opensuse-full) || rsync://stage.opensuse.org/opensuse-full/opensuse/<br/>http://mirrors.opensuse.org/list/all.html || 179 GiB || -
| OpenSUSE (opensuse-full) || rsync://stage.opensuse.org/opensuse-full/opensuse/<br/>http://mirrors.opensuse.org/list/all.html || 179 GiB || -
| Slackware || rsync://slackware.cs.utah.edu/slackware/<br/>http://slackware.com/getslack/ || 139 GiB || '''We should get added as an official mirror.'''
| Slackware || rsync://slackware.cs.utah.edu/slackware/<br/>http://slackware.com/getslack/ || 139 GiB || '''We have asked to get added as an official mirror.'''
| Ubuntu || rsync://archives.ubuntu.com/ubuntu/<br/>https://launchpad.net/ubuntu/+archivemirrors || 319 GiB || '''We used to be the official Canadian mirror (''i.e.'', ca.archive.ubuntu.com); when we get more bandwidth get us added back.'''
| Ubuntu || rsync://archives.ubuntu.com/ubuntu/<br/>https://launchpad.net/ubuntu/+archivemirrors || 319 GiB || '''We used to be the official Canadian mirror (''i.e.'', ca.archive.ubuntu.com); when we get more bandwidth get us added back.'''

Revision as of 14:56, 5 April 2010

We currently run a public mirror (mirror.csclub.uwaterloo.ca) on sodium-benzoate. We are listed on the ResNet "don't count" list so downloading from our mirror will not count against one's ResNet quota. Requests to mirror a particular distribution or archive should be made to syscom@csclub.uwaterloo.ca. We also have a bandwidth graph you can look at.

Archives Mirrored

Total Size: 4.5 TiB

Archive Upstream/Listing Size Notes
Apache rsync://rsync.us.apache.org/apache-dist/
24 GiB -
Arch Linux rsync://mirror.rit.edu/archlinux
40 GiB -
Blastwave rsync://master.rsync.blastwave.org/blastwave/
12 GiB -
CentOS rsync://us-msync.centos.org/CentOS/
118 GiB -
CPAN rsync://rsync.nic.funet.fi/CPAN/
7.3 GiB -
CRAN rsync://cran.r-project.org/CRAN/ 44 GiB We should get added as an official mirror.
CTAN rsync://carroll.aset.psu.edu/ctan/
17 GiB -
Cygwin rsync://cygwin.com/cygwin-ftp/
8.3 GiB -
Damn Small Linux rsync://ftp.heanet.ie/mirrors/damnsmalllinux.org/ 18 GiB Sent a request to be added. Maybe contact the dude listed here: http://www.damnsmalllinux.org/donate.html.
Debian rsync://ftp3.nrc.ca/debian/
489 GiB Requested to be added to the ftp.ca.debian.org rotation; will be added if the need arises..
Debian-backports rsync://www.backports.org/backports.org/
35 GiB Submitted a request to be added to the mirror list long ago with no response.
Debian-cd rsync://cdimage.debian.org/debian-cd/ 70 GiB Mirror only the first CD and DVD, and all small CD's (netinst, business-card, etc...).
Debian-multimedia rsync://www.debian-multimedia.org/debian/
6.2 GiB -
Debian-ports rsync://ftp.debian-ports.org/debian/ 44 GiB -
Debian-security rsync://security.debian.org/debian-security/ 44 GiB Debian does not currently list debian-security mirrors, and encourages users to use security.debian.org exclusively.
Debian-unofficial rsync://debian-maintainers.org/unofficial/
465 MiB -
Debian-volatile rsync://volatile-master.debian.org/debian-volatile/
3.4 GiB -
Eclipse rsync://download.eclipse.org/eclipseMirror
295 GiB -
Emdebian rsync://www.emdebian.org/debian/ 3.7 GiB Emdebian does not currently list mirrors.
FreeBSD rsync://ftp1.ca.freebsd.org/
1.2 TiB We should get added as an official mirror.
Gentoo (portage) rsync://rsync1.us.gentoo.org/gentoo-portage/ 594 MiB We are rsync4.ca.gentoo.org.
Gentoo (sources) rsync://masterdistfiles.gentoo.org/gentoo/
169 GiB See ~sysadmin/passwords/gentoo for rsync password.
GNOME rsync://ftp.gnome.org/gnome/ 91 GiB -
GNU rsync://ftp.ibiblio.org/pub/gnu/ftp/gnu/
21 GiB -
KDE rsync://master.kde.org/kdeftp/
77 GiB -
kernel.org rsync://kernel.org/pub/linux/
136 GiB -
Linux Mint (releases) rsync://ftp.heanet.ie/pub/linuxmint.com/
31 GiB -
Linux Mint (packages) rsync://packages.linuxmint.com/packages/
4.4 GiB -
mozdev rsync://rsync.mozdev.org/mozdev/
6.2 GiB Currently in the North American rotation, but could request to be added to the global rotation.
mozilla.org rsync://releases-rsync.mozilla.org/mozilla-releases/
122 GiB -
MySQL rsync://mysql.he.net/mysql/
167 GiB -
non-GNU rsync://dl.sv.gnu.org/releases/
16 GiB -
Openoffice (extended set) rsync://rsync.services.openoffice.org/openoffice-extended/
103 GiB -
OpenSUSE (opensuse-full) rsync://stage.opensuse.org/opensuse-full/opensuse/
179 GiB -
Slackware rsync://slackware.cs.utah.edu/slackware/
139 GiB We have asked to get added as an official mirror.
Ubuntu rsync://archives.ubuntu.com/ubuntu/
319 GiB We used to be the official Canadian mirror (i.e., ca.archive.ubuntu.com); when we get more bandwidth get us added back.
Ubuntu-ports rsync://ports.ubuntu.com/ubuntu-ports/ 442 GiB -
Ubuntu-ports-releases rsync://cdimage.ubuntu.com/cdimage/ubuntu-ports/releases/ 36 GiB -
Ubuntu-releases rsync://rsync.releases.ubuntu.com/releases/
39 GiB Are the official Canadian mirror (i.e., ca.releases.ubuntu.com); ubuntu-releases includes Ubuntu, Kubuntu, and Edubuntu.
xorg.freedesktop.org rsync://xorg.freedesktop.org/xorg-archive/
5.5 GiB -
Xubuntu-releases rsync://cdimage.ubuntu.com/cdimage/xubuntu/releases/
21 GiB -

Proposed Archives to Mirror

  • Fedora
  • Mandriva
  • OpenBSD (inc. OpenSSH)
  • NetBSD
  • PCLinuxOS
  • OpenSSL
  • RubyForge
  • SourceForge
  • MacPorts
  • PLT Scheme (they don't ask for mirrors, but they currently offer download from half a dozen or so sites)
  • VLC

Implementation Details

The mirroring is done by one of three scripts. The latter two are based on anonftpsync. merlin is used to call one of these scripts. Most of the scripts and such used to maintain the mirror are available in the public mirror.git repository.


ftpsync is the official Debian mirror synchronization tool, and is used to rsync the Debian repository. It's located in ~mirror/debian. Its invocation takes a few steps (this is more or less how merlin invokes it:

export BASEDIR=/home/mirror/debian
./bin/ftpsync sync:stage1
./bin/ftpsync sync:stage2


This is used to sync debian-style repositories. Its usage is:

csc-sync-debian local_dir rsync_host rsync_dir [trace_host [trace_dir]]

If trace_host is specified, then $rsync_dir/project/trace/$trace_host is checked to see if it has changed. If it has, a normal debian-style (two-pass) rsync is done.


This is used to sync a tree in a general way. Like anonftpsync, it supports locking and logging. Its usage is:

csc-sync-standard local_dir rsync_host rsync_dir


The synchronization process is run by a Python script called "merlin", written by a2brenna, stored in ~mirror/merlin. The repository list, sync time, etc. is maintained in merlin.py.


We use Apache as our web server. Here's a snippet of the worker configuration:

<IfModule mpm_worker_module>
    ServerLimit          64
    ThreadLimit          64
    StartServers          2
    MaxClients         4096
    MinSpareThreads      16
    MaxSpareThreads      48
    ThreadsPerChild      64
    MaxRequestsPerChild   0

We use the bwbar application to display current bandwidth in the footer of mirror pages.


An index of the archives we mirror is available at http://mirror.csclub.uwaterloo.ca/. As of Winter 2010, it is now generated by a Python script in ~mirror/mirror-index.

~mirror/mirror-index/make-index.py is scheduled in mirror's crontab to be run at 5:40 AM on the 14th and 28th of each month. The script can be run manually when needed (for example, when an archive is removed) as follows:

sudo -u mirror /home/mirror/mirror-index/make-index.py

This causes an instance of du to be run which computes the size of each directory. This list is then sorted alphabetically by directory name and returned to the Python script. If any errors occur during this process, the script conservatively chooses to exit rather than risk generating an index file that is incorrect.

make-index.py is configured by means of a YAML file, config.yaml, in the same directory. Its format is as follows:

docroot: /mirror/root
duflags: --human-readable --max-depth=1
output: /mirror/root/index.html

    site: apache.org
    url: http://www.apache.org/

    site: archlinux.org
    url: http://www.archlinux.org/

# (...)

The docroot is the directory which is to be scanned; this will probably always be the mirror root from which Apache serves. duflags specifies the flags to be passed to du. This is here so that it's easy to find and alter. For instance, we could change --human-readable to --si if we ever decided that, like hard disk manufacturers, we want sizes to appear larger than they are. output defines the file to which the generated index will be written.

Finally, directories specifies the list of directories to be listed. No directories not listed here will be shown. If you add a new archive and it doesn't appear, that's why. The format is fairly straightforward: simply name the directory and provide a site (the display name in the "Project Site" column) and URL.

One caveat here is that YAML does not allow tabs for whitespace. Indent with two spaces to remain consistent with the existing file format, please. Also note that the directory name is case-sensitive, as is always the case on Unix.

Finally, the HTML index file is generated from index.mako, a Mako template (which is mostly HTML anyhow). If you really can't figure out how it works, look up the Mako documentation.


We use proftpd (standalone daemon) as our ftp server. To increase performance we disable DNS lookups in proftpd.conf:

UseReverseDNS           off
IdentLookups            off

We also limit the amount of CPU/memory resources used (e.g. to minimize Globbing resources):

RLimitCPU               session 10
RLimitMemory            session 4096K

We allow a maximum of 200 concurrent ftp sessions:

MaxInstances            500
MaxClients              500


We use rsyncd (standalone daemon). We disable compression and checksumming in rsyncd.conf:

dont compress = *
refuse options = c delete

For ftp and rsync, the contents of /mirror/root/include/motd.msg are displayed when users connect.