Path: senator-bedfellow.mit.edu!senator-bedfellow.mit.edu!dreaderd!not-for-mail
From: [email protected] (Paul Blackburn)
Newsgroups: alt.filesystems.afs,alt.answers,news.answers
Subject: AFS distributed filesystem FAQ
Supersedes: <[email protected]>
Followup-To: alt.filesystems.afs
Date: Thu, 11 Jan 2018 00:03:17 -0500 (EST)
Organization: AIX Systems Support Centre, IBM UK
Sender: [email protected]
Approved: [email protected]
Expires: 24 Feb 2018 05:03:15 GMT
Message-ID: <[email protected]>
Reply-To: [email protected] (AFS FAQ comments address)
Summary: Introduction to AFS with pointers to further information
X-Last-Updated: 1998/07/10
Lines: 2908
NNTP-Posting-Host: xvm-75.mit.edu
X-Trace: 1515646998 senator-bedfellow.mit.edu 15456 18.181.2.75:55340
Xref: senator-bedfellow.mit.edu alt.filesystems.afs:2800 alt.answers:85591 news.answers:338847

Archive-name: afs-faq
Version: 1.113
Last-modified: 1950 Thursday 9th July 1998

AFS frequently asked questions
______________________________________________________________________________

  This posting contains answers to frequently asked questions about AFS.
  Your comments and contributions are welcome (email: [email protected])

  Most newsreaders can skip from topic to topic with control-G.
U  URLs: file:///afs/transarc.com/public/afs-contrib/doc/faq/afs-faq.html
N        ftp://ftp.transarc.com/pub/afs-contrib/doc/faq/afs-faq.html
N        http://www.angelfire.com/hi/plutonic/afs-faq.html
______________________________________________________________________________
Subject: Table of Contents:

  0  Preamble
     0.01  Purpose and Audience
     0.02  Acknowledgements
     0.03  Disclaimer
     0.04  Release Notes
     0.05  Quote

  1  General
     1.01  What is AFS?
     1.02  Who supplies AFS?
     1.03  What is /afs?
     1.04  What is an AFS cell?
     1.05  What are the benefits of using AFS?
           1.05.a  Cache Manager
           1.05.b  Location independence
           1.05.c  Scalability
           1.05.d  Improved security
           1.05.e  Single systems image (SSI)
           1.05.f  Replicated AFS volumes
           1.05.g  Improved robustness to server crash
           1.05.h  "Easy to use" networking
           1.05.i  Communications protocol
           1.05.j  Improved system management capability
U     1.06  Which systems is AFS available for?
U     1.07  What does "ls /afs" display in the Internet AFS filetree?
     1.08  Why does AFS use Kerberos authentication?
     1.09  Does AFS work over protocols other than TCP/IP?
     1.10  How can I access AFS from my PC?
     1.11  How does AFS compare with NFS?

  2  Using AFS
     2.01  What are the differences between AFS and a unix filesystem?
     2.02  What is an AFS protection group?
     2.03  What are the AFS defined protection groups?
     2.04  What is an AFS access control list (ACL)?
     2.05  What are the AFS access rights?
     2.06  What is pagsh?
     2.07  Why use a PAG?
     2.08  How can I tell if I have a PAG?
     2.09  Can I still run cron jobs with AFS?
     2.10  How much disk space does a 1 byte file occupy in AFS?
     2.11  Is it possible to specify a user who is external
           to the current AFS cell on an ACL?
     2.12  Are there any problems printing files in /afs?
     2.13  Can I create a fifo (aka named pipe) in /afs?
     2.14  If an AFS server crashes, do I have to reboot my AFS client?
     2.15  Can I use AFS on my diskless workstation?
     2.16  Can I test for AFS tokens from within my program?
     2.17  What's the difference between /afs/cellname and /afs/.cellname?
     2.18  Can I klog as two users on one machine in the same cell?
     2.19  What are the ~/.__afsXXXX files?

  3  AFS administration
     3.01  Is there a version of xdm available with AFS authentication?
     3.02  Is there a version of xlock available with AFS authentication?
     3.03  What is /afs/@cell?
     3.04  Given that AFS data is location independent, how does
           an AFS client determine which server houses the data
           its user is attempting to access?
     3.05  Which protocols does AFS use?
     3.06  Are setuid programs executable across AFS cell boundaries?
     3.07  How does AFS maintain consistency on read-write files?
     3.08  How can I run daemons with tokens that do not expire?
     3.09  Can I check my user's passwords for security purposes?
     3.10  Is there a way to automatically balance disk usage across
           fileservers?
     3.11  Can I shutdown an AFS fileserver without affecting users?
     3.12  How can I set up mail delivery to users with $HOMEs in AFS?
     3.13  Should I replicate a ReadOnly volume on the same partition
           and server as the ReadWrite volume?
     3.14  Should I start AFS before NFS in /etc/inittab?
     3.15  Will AFS run on a multi-homed fileserver?
     3.16  Can I replicate my user's home directory AFS volumes?
     3.17  Which TCP/IP ports and protocols do I need to enable
           in order to operate AFS through my Internet firewall?
     3.18  What is the Andrew Benchmark?
U     3.19  Is there a version of HP VUE login with AFS authentication?
     3.20  How can I list which clients have cached files from a server?
     3.21  Do Backup volumes require as much space as ReadWrite volumes?
     3.22  Should I run timed on my AFS client?
     3.23  Why should I keep /usr/vice/etc/CellServDB current?
     3.24  How can I keep /usr/vice/etc/CellServDB current?
     3.25  How can I compute a list of AFS fileservers?
     3.26  How can I set up anonymous FTP login to access /afs?
     3.27  Where can I find the Andrew Benchmark?

  4  Getting more information
     4.01  Is there an anonymous FTP site with AFS information?
     4.02  Which USENET newsgroups discuss AFS?
     4.03  Where can I get training in AFS?
U     4.04  Where can I find AFS resources in World Wide Web (WWW)?
     4.05  Is there a mailing list for AFS topics?
U     4.06  Where can I find an archive of [email protected]?
     4.07  Where can I find an archive of alt.filesystems.afs?
U     4.08  Where can I find AFS related GIFs?
     4.09  Gibt es eine deutsche AFS Benutzer Gruppe?
     4.10  Donde puedo encontrar informacion en Espanol sobre AFS?

  5  About the AFS faq
U     5.01  How can I get a copy of the AFS faq?
     5.02  How can I get my question (and answer) into the AFS faq?
U     5.03  How can I access the AFS faq via the World Wide Web?

  6  Bibliography

  7  Change History
______________________________________________________________________________

Subject: 0  Preamble

Subject: 0.01  Purpose and audience

  The aim of this compilation is to provide information about AFS including:

     + A brief introduction
     + Answers to some often asked questions
     + Pointers to further information

  Definitive and detailed information on AFS is provided in Transarc's
  AFS manuals ([23], [24], [25]).

  The intended audience ranges from people who know little of the subject
  and want to know more to those who have experience with AFS and wish
  to share useful information by contributing to the faq.

Subject: 0.02  Acknowledgements

  The information presented here has been gleaned from many sources.
  Some material has been directly contributed by people listed below.

  I would like to thank the following for contributing:

       Pierette Maniago VanRyzin (Transarc)
       Lyle Seaman (Transarc)
       Joseph Jackson (Transarc)
       Dan Lovinger (Microsoft)
       Lucien Van Elsen (IBM)
       Jim Rees (University of Michigan)
       Derrick J. Brashear (Carnegie Mellon University)
       Hans-Werner Paulsen (MPI fuer Astrophysik, Garching)
       Margo Hikida (Hewlett Packard)
       Michael Fagan (IBM)
       Robert Malick (National Institute of Health, USA)
       Rainer Toebbicke (European Laboratory for Particle Physics, CERN)
       Mic Bowman (Transarc)
       Mike Prince (IBM)
       Bob Oesterlin (IBM)
       Pat Wilson (Dartmouth College)
       Cristian Espinoza (Pontificia Universidad Catolica de Chile)
       Mary Ann DelBusso (Transarc)
       Michael Niksch (IBM)
N       Kelly Chambers (Transarc)

  Thanks also to indirect contributors:

       Ken Paquette (IBM)
       Lance Pickup (IBM)
       Lisa Chavez (IBM)
       Dawn E. Johnson (Transarc)
       David Snearline (University of Michigan Engineering)
       Rens Troost (New Century Systems)
       Anton Knaus (Carnegie Mellon University)
       Mike Shaddock (SAS Institute Inc.)

  If this compilation has any merit then much credit belongs to Pierette
  for giving inspiration, support, answers, and proof-reading.

Subject: 0.03  Disclaimer

  I make no representation about the suitability of this
  information for any purpose.

  While every effort is made to keep the information in
  this document accurate and current, it is provided "as is"
  with no warranty expressed or implied.

Subject: 0.04  Release Notes

  This compilation contains material used with permission of
  Transarc Corporation. Permission to copy is given provided any
  copyright notices and acknowledgements are retained.

  Column 1 is used to indicate changes from the last issue:

     N = new item
     U = updated item

  Changes from the last version are to be found at the end of this file.
______________________________________________________________________________
Subject: 0.05  Quote

  "'Tis true; there's magic in the web of it;"         Othello, Act 3 Scene 4
                                            --William Shakespeare (1564-1616)
______________________________________________________________________________
Subject: 1  General

Subject: 1.01  What is AFS?

  AFS is a distributed filesystem that enables co-operating hosts
  (clients and servers) to efficiently share filesystem resources
  across both local area and wide area networks.

  AFS is marketed, maintained, and extended by Transarc Corporation.

  AFS is based on a distributed file system originally developed
  at the Information Technology Center at Carnegie-Mellon University
  that was called the "Andrew File System".

  "Andrew" was the name of the research project at CMU - honouring the
  founders of the University.  Once Transarc was formed and AFS became a
  product, the "Andrew" was dropped to indicate that AFS had gone beyond
  the Andrew research project and had become a supported, product quality
  filesystem. However, there were a number of existing cells that rooted
  their filesystem as /afs. At the time, changing the root of the filesystem
  was a non-trivial undertaking. So, to save the early AFS sites from having
  to rename their filesystem, AFS remained as the name and filesystem root.

Subject: 1.02  Who supplies AFS?

       Transarc Corporation          phone: +1 (412) 338-4400
       The Gulf Tower
       707 Grant Street              fax:   +1 (412) 338-4404
       Pittsburgh
       PA 15219                      email: [email protected]
       United States of America             [email protected]

                                     WWW:    http://www.transarc.com

Subject: 1.03  What is /afs?

  The root of the AFS filetree is /afs. If you execute "ls /afs" you will
  see directories that correspond to AFS cells (see below). These cells
  may be local (on same LAN) or remote (eg halfway around the world).

  With AFS you can access all the filesystem space under /afs with commands
  you already use (eg: cd, cp, rm, and so on) provided you have been granted
  permission (see AFS ACL below).

Subject: 1.04  What is an AFS cell?

  An AFS cell is a collection of servers grouped together administratively
  and presenting a single, cohesive filesystem.  Typically, an AFS cell is
  a set of hosts that use the same Internet domain name.

  Normally, a variation of the domain name is used as the AFS cell name.

  Users log into AFS client workstations which request information and files
  from the cell's servers on behalf of the users.

Subject: 1.05  What are the benefits of using AFS?

  The main strengths of AFS are its:

     + caching facility
     + security features
     + simplicity of addressing
     + scalability
     + communications protocol

  Here are some of the advantages of using AFS in more detail:

Subject: 1.05.a  Cache Manager

  AFS client machines run a Cache Manager process. The Cache Manager
  maintains information about the identities of the users logged into
  the machine, finds and requests data on their behalf, and keeps chunks
  of retrieved files on local disk.

  The effect of this is that as soon as a remote file is accessed
  a chunk of that file gets copied to local disk and so subsequent
  accesses (warm reads) are almost as fast as to local disk and
  considerably faster than a cold read (across the network).

  Local caching also significantly reduces the amount of network traffic,
  improving performance when a cold read is necessary.

Subject: 1.05.b  Location independence

  Unlike NFS, which makes use of /etc/filesystems (on a client) to map
  (mount) between a local directory name and a remote filesystem, AFS
  does its mapping (filename to location) at the server. This has the
  tremendous advantage of making the served filespace location independent.

  Location independence means that a user does not need to know which
  fileserver holds the file, the user only needs to know the pathname
  of a file. Of course, the user does need to know the name of the
  AFS cell to which the file belongs. Use of the AFS cellname as the
  second part of the pathname (eg: /afs/$AFSCELL/somefile) is helpful
  to distinguish between file namespaces of the local and non-local
  AFS cells.

  To understand why such location independence is useful, consider
  having 20 clients and two servers. Let's say you had to move
  a filesystem "/home" from server a to server b.

  Using NFS, you would have to change the /etc/filesystems file on 20
  clients and take "/home" off-line while you moved it between servers.

  With AFS, you simply move the AFS volume(s) which constitute "/home"
  between the servers. You do this "on-line" while users are actively
  using files in "/home" with no disruption to their work.

  (Actually, the AFS equivalent of "/home" would be /afs/$AFSCELL/home
  where $AFSCELL is the AFS cellname.)

Subject: 1.05.c  Scalability

  With location independence comes scalability. An architectural goal
  of the AFS designers was client/server ratios of 200:1 which has
  been successfully exceeded at some sites.

  Transarc do not recommend customers use the 200:1 ratio. A more
  cautious value of 50:1 is expected to be practical in most cases.
  It is certainly possible to work with a ratio somewhere between
  these two values. Exactly what value depends on many factors including:
  number of AFS files, size of AFS files, rate at which changes are made,
  rate at which file are being accessed, speed of servers processor,
  I/O rates, and network bandwidth.

  AFS cells can range from the small (1 server/client) to the massive
  (with tens of servers and thousands of clients).

  Cells can be dynamic: it is simple to add new fileservers or clients
  and grow the computing resources to meet new user requirements.

Subject: 1.05.d  Improved security

  Firstly, AFS makes use of Kerberos to authenticate users.
  This improves security for several reasons:

    + passwords do not pass across the network in plaintext

    + encrypted passwords no longer need to be visible

         You don't have to use NIS, aka yellow pages, to distribute
         /etc/passwd - thus "ypcat passwd" can be eliminated.

         If you do choose to use NIS, you can replace the password
         field with "X" so the encrypted password is not visible.
         (These issues are discussed in detail in [25]).

    + AFS uses mutual authentication - both the service provider
      and service requester prove their identities

  Secondly, AFS uses access control lists (ACLs) to enable users to
  restrict access to their own directories.

Subject: 1.05.e  Single systems image (SSI)

  Establishing the same view of filestore from each client and server
  in a network of systems (that comprise an AFS cell) is an order of
  magnitude simpler with AFS than it is with, say, NFS.

  This is useful to do because it enables users to move from workstation
  to workstation and still have the same view of filestore. It also
  simplifies part of the systems management workload.

  In addition, because AFS works well over wide area networks the SSI
  is also accessible remotely.

  As an example, consider a company with two widespread divisions
  (and two AFS cells): ny.acme.com and sf.acme.com. Mr Fudd, based
  in the New York office, is visiting the San Francisco office.

  Mr. Fudd can then use any AFS client workstation in the San Francisco
  office that he can log into (a unprivileged guest account would suffice).
  He could authenticate himself to the ny.acme.com cell and securely access
  his New York filespace.

  For example:

      The following shows a guest in the sf.acme.com AFS cell:
      {0} add AFS executables directory to PATH
      {1} obtaining a PAG with pagsh command (see 2.06)
      {2} use the klog command to authenticate into the ny.acme.com AFS cell
      {3} making a HOME away from home
      {4} invoking a homely .profile

      [email protected] $ PATH=/usr/afsws/bin:$PATH       # {0}
      [email protected] $ pagsh                           # {1}
      $ klog -cell ny.acme.com -principal elmer                    # {2}
      Password:
      $ HOME=/afs/ny.acme.com/user/elmer; export HOME              # {3}
      $ cd
      $ .  .profile                                                # {4}
      you have new mail
      guest@toontown $

  It is not necessary for the San Francisco sys admin to give Mr. Fudd
  an AFS account in the sf.acme.com cell.  Mr. Fudd only needs to be
  able to log into an AFS client that is:
     1) on the same network as his cell and
     2) his ny.acme.com cell is mounted in the sf.acme.com cell
        (as would certainly be the case in a company with two cells).

Subject: 1.05.f  Replicated AFS volumes

  AFS files are stored in structures called Volumes.  These volumes
  reside on the disks of the AFS file server machines.  Volumes containing
  frequently accessed data can be read-only replicated on several servers.

  Cache managers (on users client workstations) will make use of replicate
  volumes to load balance.  If accessing data from one replicate copy, and
  that copy becomes unavailable due to server or network problems, AFS will
  automatically start accessing the same data from a different replicate copy.

  An AFS client workstation will access the closest volume copy.
  By placing replicate volumes on servers closer to clients (eg on same
  physical LAN) access to those resources is improved and network traffic
  reduced.

Subject: 1.05.g  Improved robustness to server crash

  The Cache Manager maintains local copies of remotely accessed files.

  This is accomplished in the cache by breaking files into chunks
  of up to 64k (default chunk size). So, for a large file, there may be
  several chunks in the cache but a small file will occupy a single chunk
  (which will be only as big as is needed).

  A "working set" of files that have been accessed on the client is
  established locally in the client's cache (copied from fileserver(s)).

  If a fileserver crashes, the client's locally cached file copies
  remain readable but updates to cached files fail while the server is down.

  Also, if the AFS configuration has included replicated read-only volumes
  then alternate fileservers can satisfy requests for files from those
  volumes.

Subject: 1.05.h  "Easy to use" networking

  Accessing remote file resources via the network becomes much simpler
  when using AFS. Users have much less to worry about: want to move
  a file from a remote site? Just copy it to a different part of /afs.

  Once you have wide-area AFS in place, you don't have to keep local
  copies of files. Let AFS fetch and cache those files when you need them.

Subject: 1.05.i  Communications protocol

  AFS communications protocol is optimized for Wide Area Networks.
  Retransmitting only the single bad packet in a batch of packets
  and allowing the number of unacknowledged packets to be higher
  (than in other protocols, see [4]).

Subject: 1.05.j  Improved system management capability

  Systems administrators are able to make configuration changes
  from any client in the AFS cell (it is not necessary to login
  to a fileserver).

  With AFS it is simple to effect changes without having to take
  systems off-line.

  Example:

  A department (with its own AFS cell) was relocated to another office.
  The cell had several fileservers and many clients.
  How could they move their systems without causing disruption?

  First, the network infrastructure was established to the new location.
  The AFS volumes on one fileserver were migrated to the other fileservers.
  The "freed up" fileserver was moved to the new office and connected
  to the network.

  A second fileserver was "freed up" by moving its AFS volumes across
  the network to the first fileserver at the new office. The second
  fileserver was then moved.

  This process was repeated until all the fileservers were moved.

  All this happened with users on client workstations continuing
  to use the cell's filespace. Unless a user saw a fileserver
  being physically moved (s)he would have no way to tell the change
  had taken place.

  Finally, the AFS clients were moved - this was noticed!

Subject: 1.06  Which systems is AFS available for?

  AFS runs on systems from: HP, Next, DEC, IBM, SUN, and SGI.

  Transarc customers have done ports to Crays, and the 3090, but all
  are based on some flavour of unix.  Some customers have done work to
  make AFS data available to PCs and Macs, although they are using
  something similar to the AFS/NFS translator (a system that enables
  "NFS only" clients to NFS mount the AFS filetree /afs).

  There is a client only implementation "AFS Client for Windows/NT".

N  A page describing the current systems for which AFS is supported
N  may be found at:
N
N     http://www.transarc.com/Support/afs/relversions/platforms.html
N
  There are also ports of AFS done by customers available from Transarc
  on an "as is" unsupported basis.

  More information on this can be found at:

     /afs/transarc.com/public/afs-contrib/bin/README
     ftp://ftp.transarc.com/pub/afs-contrib/bin/README

  These ports of AFS client code include:

     HP (Apollo) Domain OS - by Jim Rees at the University of Michigan.
     sun386i - by Derek Atkins and Chris Provenzano at MIT.
     Linux - by Derek Atkins, mailing list: <[email protected]>
                 http://www.mit.edu:8008/menelaus/linux-afs/
     NetBSD - by John Kohl, mailing list: <[email protected]>

  There is some information about AFS on OS/2 at:
     http://www.club.cc.cmu.edu/~jgrande/afsos2.html

N  The AFS on Linux FAQ may be found at:
N     http://www.umlug.umd.edu/linuxafs/

Subject: 1.07  What does "ls /afs" display in the Internet AFS filetree?

  Essentially this displays the AFS cells that co-operate in the
  Internet AFS filetree.

  Note that the output of this will depend on the cell you do it from;
  a given cell may not have all the publicly advertised cells available,
  and it may have some cells that aren't advertised outside of the given site.

  The definitive source for this information is:

     file:///afs/transarc.com/service/etc/CellServDB.export

  I've included the list of cell names included in it below:

  asu.edu                 #ASU
  uni-freiburg.de         #Albert-Ludwigs-Universitat Freiburg
  anl.gov                 #Argonne National Laboratory
  fl.mcs.anl.gov          # Argonne National Laboratory MCS Division FL
  dapnia.saclay.cea.fr    #Axlan-CEA
  bcc.ac.uk               #Bloomsbury Computing Consortium
  bu.edu                  #Boston University
  cs.brown.edu            #Brown University Department of Computer Science
  caspur.it               #CASPUR Inter-University Computing Consortium,Rome
  ciesin.org              #CIESIN
  mathematik-cip.uni-stuttgart.de #CIP-Pool of Math. Dept, Univ. Stuttgart
  gg.caltech.edu          #Caltech Computer Graphics Group
  cards.com               #Cards - Electronic Warfare Associates
  cheme.cmu.edu           #Carnegie Mellon Univ. Chemical Engineering Dept.
  cmu.edu                 #Carnegie Mellon University
  andrew.cmu.edu          #Carnegie Mellon University - Campus
  ce.cmu.edu              #Carnegie Mellon University - Civil Eng. Dept.
  ece.cmu.edu             #Carnegie Mellon University - Elec. Comp. Eng. Dept.
  me.cmu.edu              #Carnegie Mellon University - Mechanical Engineering
  cs.cmu.edu              #Carnegie Mellon University - School of Comp. Sci.
  club.cc.cmu.edu         #Carnegie Mellon University Computer Club
  cert.org                #CERT/Coordination Center
  others.chalmers.se      #Chalmers University of Technology - General users
  cipool.uni-stuttgart.de #CIP Pool, Rechenzentrum University of Stuttgart
  clarkson.edu            #Clarkson University, Potsdam, USA
  msc.cornell.edu         #Cornell University Materials Science Center
  graphics.cornell.edu    #Cornell University Program of Computer Graphics
  theory.cornell.edu      #Cornell University Theory Center
  ifh.de                  #DESY-IfH Zeuthen
  northstar.dartmouth.edu #Dartmouth College, Project Northstar
  desy.de                 #Deutsches Elektronen-Synchrotron
  dkrz.de                 #Deutsches Klimarechenzentrum Hamburg
  dis.uniroma1.it         #DIS, Univ. "La Sapienza", Rome, area Buonarotti
  msrc.pnl.gov            #EMSL's AFS Cell
  zdvpool.uni-tuebingen.de#Eberhard-Karls-Universitaet Tuebingen, WS-Pools
  enea.it                 #enea.it
  es.net                  #Energy Sciences Net
  research.ec.org         #Esprit Research Network of Excellence
  dce.emsl.pnl.gov        #EMSL's DCE Cell
  cern.ch                 #European Laboratory for Particle Physics, Geneva
  fnal.gov                #Fermi National Acclerator Laboratory
  fh-heilbronn.de         #Fachhochschule Heilbronn
  hephy.at                #hephy-vienna
  sleeper.nsa.hp.com      #HP Cupertino
  palo_alto.hpl.hp.com    #HP Palo Alto
  afs.hursley.ibm.com     #IBM Hursley Laboratories (UK), external cell
  ibm.uk                  #IBM UK, AIX Systems Support Centre
  zurich.ibm.ch           #IBM Zurich Internet Cell
  ctp.se.ibm.com          #IBM/4C, Chalmers, Sweden
  ipp-hgw.mpg.de          #IPP site at Greifswald
  in2p3.fr                #IN2P3 production cell
  lngs.infn.it            #INFN Laboratori Nazionali di Gran Sasso, Italia
  le.infn.it              #INFN Sezione di Lecce, Italia
  pi.infn.it              #INFN Sezione di Pisa
  ike.uni-stuttgart.de    #Institut fuer Kernenergetik, Universitaet Stuttgart
  ipp-garching.mpg.de     #Institut fuer Plasmaphysik
  csv.ica.uni-stuttgart.de #Institut fuer Computeranwendungen, Uni. Stuttgart
  iastate.edu             #Iowa State University
  infn.it                 #Istituto Nazionale di Fisica Nucleare, Italia
  jpl.nasa.gov            #Jet Propulsion Laboratory
  zdv.uni-mainz.de        #Johannes-Gutenberg-Universitaet Mainz
  isk.kth.se              #KTH College of Engineering
  cc.keio.ac.jp           #Keio University, Fac. of Sci. & Tech. Computing Ctr
  sfc.keio.ac.jp          #Keio University, Japan
  afs-math.zib-berlin.de  #Konrad-Zuse-Zentrum fuer Informationstechnik Berlin
  thermo-a.mw.tu-muenchen.de #Lehrstuhl A fuer Thermodynamik,TUM
  lrz-muenchen.de         #Leibniz-Rechenzentrum Muenchen Germany
  athena.mit.edu          #MIT/Athena cell
  net.mit.edu             #MIT/Network Group cell
  sipb.mit.edu            #MIT/SIPB cell
  msu.edu                 #Michigan State University home cell
  mpa-garching.mpg.de     #Max-Planck-Institut fuer Astrophysik
  federation.atd.net      #Multi Resident AFS at Naval Research Lab - CCS
  isl.ntt.jp              #NTT Information and Communication
  nersc.gov               #National Energy Research Supercomputer Center
  alw.nih.gov             #National Institutes of Health
  nrel.gov                #National Renewable Energy Laboratory
  cmf.nrl.navy.mil        #Naval Research Lab
  lcp.nrl.navy.mil        #Naval Research Lab - Lab for Computational Physics
  nrlfs1.nrl.navy.mil     #Naval Research Laboratory
  eos.ncsu.edu            #NCSU - College of Engineering
  unity.ncsu.edu          #NCSU Campus
  ncat.edu                #North Carolina Agricultural and Technical State U.
  bp.ncsu.edu             #North Carolina State University - Backbone Prototype
  ri.osf.org              #OSF Research Institute
  gr.osf.org              #OSF Research Institute, Grenoble
  urz.uni-magdeburg.de    #Otto-von-Guericke-Universitaet, Magdeburg
N  ovpit.indiana.edu       #OVPIT at Indiana University
  psc.edu                 #PSC (Pittsburgh Supercomputing Center)
  psu.edu                 #Penn State
  phy.bnl.gov             #Physics Deptpartment, Brookhaven National Lab
  postech.ac.kr           #Pohang University of Science
  pppl.gov                #Princeton Plasma Physics Laboratory
  rwcp.or.jp              #Real World Computer Partnership(rwcp)
  rz.uni-jena.de          #Rechenzentrum University of Jena, Germany
  rhrk.uni-kl.de          #Rechenzentrum University of Kaiserslautern
  rus.uni-stuttgart.de    #Rechenzentrum University of Stuttgart
  rhic                    #Relativistic Heavy Ion Collider
  rpi.edu                 #Rensselaer Polytechnic Institute
  uni-bonn.de             #Rheinische Friedrich Wilhelm Univesitaet Bonn
  rose-hulman.edu         #Rose-Hulman Institute of Technology
  cs.rose-hulman.edu      # Rose-Hulman Inst. of Tech., CS Department
  nada.kth.se             #Royal Institute of Technology, NADA
  rl.ac.uk                #Rutherford Appleton Lab, England
  slac.stanford.edu       #Stanford Linear Accelerator Center
  dsg.stanford.edu        #Stanford Univ. - Comp. Sci. - Distributed Systems
  ir.stanford.edu         #Stanford University
  afs1.scri.fsu.edu       #Supercomputer Computations Research Instit
  ethz.ch                 #Swiss Federal Inst. of Tech. - Zurich, Switzerland
  hrzone.th-darmstadt.de  #TH-Darmstadt
  tu-bs.de                #Technical University of Braunschweig, Germany
  tu-chemnitz.de          #Technische Universitaet Chemnitz-Zwickau, Germany
  telos.com               #Telos Systems Group - Chantilly, Va.
  transarc.com            #Transarc Corporation
  cats.ucsc.edu           #UC Santa Cruz, Comp and Tech Services, California
  umr.edu                 #UMR - Missouri's Technological University
  hep.net                 #US High Energy Physics Information cell
  uni-mannheim.de         #Uni Mannheim (Rechenzentrum)
  ece.ucdavis.edu         #Univ California - Davis campus
  geo.uni-koeln.de        #Univ. of Cologne Inst. for Geophysics & Meteorology
  meteo.uni-koeln.de      #Univ. of Cologne Inst. for Geophysics & Meteorology
N  dsi.uniroma1.it         #Univ. Rome-1, Dept. of Computer Science
U  spv.uniroma1.it         #Univ. Rome-1, Area San Pietro in Vincoli
N  vn.uniroma3.it          #Univ. Rome-3, Area Vasca Navale
  urz.uni-heidelberg.de   #Universitaet Heidelberg
  spc.uchicago.edu        #University of Chicago - Social Sciences
  rrz.uni-koeln.de        #University of Cologne -  Reg Comp Center
  wu-wien.ac.at           #University of Economics, Vienna, Austria
  uni-hohenheim.de        #University of Hohenheim
  ncsa.uiuc.edu           #University of Illinois
  wam.umd.edu             #University of Maryland Network WAM Project
  glue.umd.edu            #University of Maryland - Project Glue
  engin.umich.edu         #University of Michigan - CAEN
  umich.edu               #University of Michigan - Campus
  dmsv.med.umich.edu      #University of Michigan - DMSV
  citi.umich.edu          #University of Michigan - IFS Development
  lsa.umich.edu           #University of Michigan - LSA College
  math.lsa.umich.edu      #University of Michigan - Math Cell
  sph.umich.edu           #University of Michigan -- School of Public
  cs.unc.edu              #University of North Carolina at Chapel Hill
  nd.edu                  #University of Notre Dame
  pitt.edu                #University of Pittsburgh
  vn.uniroma3.it          #University of Rome 3, Area Vasca Navale, Italy
  isi.edu                 #University of Southern California/ISI
  dce.uni-stuttgart.de    #University of Stuttgart - DCE/DFS Cell
  ihf.uni-stuttgart.de    #University of Stuttgart, Ins. fuer Hochfrequenz-Tec
  mathematik.uni-stuttgart.de #University of Stuttgart, Math Dept.
  cs.utah.edu             #University of Utah Computer Science Dept
  utah.edu                #University of Utah Information Tech. Service
  cs.washington.edu       #University of Washington Comp Sci Department
  wisc.edu                #University of Wisconsin-Madison, Campus
  cs.wisc.edu             #University of Wisconsin-Madison, Comp Sci Dept
  belwue.uni-tuebingen.de #ZDV Universitaet Tuebingen

  This shows different and widespread organizations making use
  of the Internet AFS filetree.

  Note that it is also possible to use AFS "behind the firewall"
  within the confines of your organization's network - you don't have
  to participate in the Internet AFS filetree.

  Indeed, there are lots of benefits of using AFS on a local area network
  without using the WAN capabilities.

Subject: 1.08  Why does AFS use Kerberos authentication?

  It improves security.

  Kerberos uses the idea of a trusted third party to prove identification.
  This is a bit like using a letter of introduction or quoting a referee
  who will vouch for you.

  When a user authenticates using the klog command (s)he is prompted
  for a password. If the password is accepted the Kerberos
  Authentication Server (KAS) provides the user with an encrypted token
  (containing a "ticket granting ticket").

  From that point on, it is the encrypted token that is used to prove
  the user's identity. These tokens have a limited lifetime (typically
  a day) and are useless when expired.

  In AFS, it is possible to authenticate into multiple AFS cells.
  A summary of the current set of tokens held can be displayed
  by using the "tokens" command.

  For example:
     elmer@toontown $ tokens

     Tokens held by the Cache Manager:

     User's (AFS ID 9997) tokens for [email protected] [Expires Sep 15 06:50]
     User's (AFS ID 5391) tokens for [email protected] [Expires Sep 15 06:48]
        --End of list--

  Kerberos improves security because a users's password need only be
  entered once (at klog time).

  AFS uses Kerberos to do complex mutual authentication which means that
  both the service requester and the service provider have to prove their
  identities before a service is granted.

  Transarc's implementation of Kerberos is slightly different from
  MIT Kerberos V4 but AFS can work with either version.

  Joe Jackson  wrote about this in:
  http://www.cs.cmu.edu/afs/andrew.cmu.edu/usr/shadow/www/afs/afs-with-kerberos.html

  For more detail on this and other Kerberos issues see the faq
  for Kerberos (posted to news.answers and comp.protocols.kerberos) [28].
  (Also, see [15], [16], [26], [27])

Subject: 1.09  Does AFS work over protocols other than TCP/IP?

  No. AFS was designed to work over TCP/IP.

Subject: 1.10  How can I access AFS from my PC?

  You can use PC-Interface which is available from Transarc and
  Locus Computing Corporations.

  For more information on PC-Interface see the PC-Interface
  Frequently Asked Questions file in:

     file:///afs/transarc.com/public/afs-contrib/doc/faq/pci.faq
     ftp://ftp.transarc.com/pub/afs-contrib/doc/faq/pci.faq

  There is also SAMBA (an SMB/netbios server for UNIX). The current
  version will authenticate the connecting process with AFS as well.

U     http://samba.anu.edu.au/samba/

  The SAMBA FAQ is in:

U     http://samba.anu.edu.au/samba/docs/faq/sambafaq-1.html#ss1.1

  The SAMBA mailing list can be joined via: [email protected]

Subject: 1.11  How does AFS compare with NFS?

                               AFS                          NFS
  File Access           Common name space from     Different file names from
                        all workstations           different workstations

  File Location         Automatic tracking by      Mountpoints to files set by
  Tracking              file system processes      administrators and users
                        and databases

  Performance           Client caching to reduce   No local disk caching;
                        network load; callbacks    limited cache consistency
                        to maintain cache consis-
                        tency

  Andrew Benchmark      Average time of 210        Average time of 280
  (5 phases, 8 clients) seconds/client             seconds/client

  Scaling capabilities  Maintains performance in   Best in small to mid-size
                        small and very large       installations
                        installations

                        Excellent performance on   Best in local-area
                        wide-area configuration    configurations

  Security              Kerberos mutual authen-    Security based on
                        tication                   unencrypted user ID's

                        Access control lists on    No access control lists
                        directories for user and
                        group access

  Availability          Replicates read-mostly     No replication
                        data and AFS system
                        information

  Backup Operation      No system downtime with    Standard UNIX backup system
                        specially developed AFS
                        Backup System

  Reconfiguration       By volumes (groups of      Per-file movement
                        files)

                        No user impact; files      Users lose access to files
                        remain accessible during   and filenames change
                        moves, and file names do   (mountpoints need to be
                        not change                 reset)

  System Management     Most tasks performed from  Frequently involves telnet
                        any workstation            to other workstations

  Autonomous            Autonomous administrative  File servers and clients
  Architecture          units called cells, in
                        addition to file servers
                        and clients

                        No trust required between  No security distinctions
                        cells                      between sites

  [ source: ftp://ftp.transarc.com/pub/afsps/doc/afs-nfs.comparison ]

  Other points:

     + Some vendors offer more secure versions of NFS but implementations vary.
       Many NFS ports have no extra security features (such as Kerberos).

     + The AFS Cache Manager can be configured to work with a RAM (memory)
       based cache. This offers signifigant performance benefits over
       a disk based cache.

       NFS has no such feature.
       Imagine how much faster it is to access files cached into RAM!

     + The Andrew benchmark demonstrates that AFS has better performance
       than NFS as the number of clients increases. A graph of this
       (taken from Andrew benchmark report) is available in:

U       http://www.angelfire.com/hi/plutonic/images/andrew1.jpg

Subject: 2  Using AFS

Subject: 2.01  What are the differences between AFS and a unix filesystem?

  Essentially, from a user's point of view, there is little difference
  between AFS and local unix filestore. Nearly all the commands normally
  used to access local files can be used to access files in /afs.

  In the following set of sections, I have attempted to "target"
  each section to an appropriate type of user by including to the
  right of each section heading one of: User, Programmer, SysAdmin.

  Here is a summary of the differences:

  Authentication:                                         [ User ]

     Before a user can access protected AFS files (s)he needs to become
     authenticated to AFS using the klog command (Kerberos login) to get
     a Kerberos "ticket granting ticket" (called a token from here on).

     Without a token, an unauthenticated user is given the AFS identity
     "system:anyuser" and as such is only able to access files in directories
     that have ACLs granting system:anyuser access.

     Many systems have the klog function built into the system login program.
     So a user would not even have to know they gain a token on logging in.
     If you use a system where you have to issue the klog command after
     login then you should run the pagsh command first (see below).

     AFS provides access control lists to give more precise control
     to users wishing to protect their files (see AFS ACL below).

  File permissions:                                       [ User ]

     Unix mode bits for group and other are ignored.
     The mode bits for the file owner don't work the way they used to.

     Users should protect their AFS files with (directory) ACLs only.
     Just use mode bits to make a file executable.

  Data protection with AFS ACLs:                          [ User ]

     Some versions of unix (eg IBM's AIX version 3) allow ACLs on
     local files. In AFS, ACLs protect directories and used with
     AFS protection groups (see below) provide a finer granularity
     of protection than can be achieved with basic unix file permissions.
     (AFS ACLs are described in more detail below.)

  Protection groups:                                      [ User ]

     Users can create and maintain their own protection groups in AFS -
     as opposed to unix where only sys admins can manage protection groups.

  Hard links:                                             [ User ]

     In AFS, hard links (eg: ln old new) are only valid within a directory.
     This is because AFS ACLs protect directories (not individual files)
     and allowing hard links that span directories would subvert ACL
     protection.

     Symbolic links work in AFS because they reference a pathname and
     not an i-node directly. (Hard links reference an i-node directly.)

  Changing file protection by moving a file:              [ User ]

     Moving a file to a different directory will change the protection
     of a file if the ACL on the new directory if different to the ACL
     on the original directory.

  chown and chgrp:                                        [ User ]

     Only members of the AFS group "system:administrators" can use these
     commands on files in /afs.

  Save on close:                                          [ Programmer ]

     AFS Cache Manager does not send file modifications to a file server
     until the close() or fsync() system call.

     write() system calls only update the local cache copy on the client.

     Note the difference in semantic of writing a file:

     local unix file: writes update the file "immediately"
     AFS file:        local cached copy updated "immediately" but
                      the server copy is only updated when the file
                      is closed or fsync'ed.

     It is important to understand that most applications (eg: vi, emacs,
     frame, interleaf, wingz, dogz, etc) issue the close() system call when
     the user chooses/issues the "save" command in the application.

     Users are not required to exit the application to "save" their
     changes back to the server.

  byte-range file locking:                                [ Programmer ]

     AFS does not support byte-range locking within a file,
     although lockf() and fcntl() calls will return 0 (success).
     The first time a byte-range lock is attempted, AFS will display:

     "afs: byte-range lock/unlock ignored; make sure no one else
      else is running this program."

  whole file locking:                                     [ Programmer ]

     AFS does support advisory locking an entire file with flock().
     Processes on the same client workstation that attempt to lock
     a file obey the proper locking semantics.

     Processes on different AFS clients requesting a lock on the same
     file would get EWOULDBLOCK returned.

  character and block special files:                      [ SysAdmin ]

     AFS does not support character and block special files.
     The mknod command does not create either character or block
     special files in /afs.

  AFS version of fsck:                                    [ SysAdmin ]

     On an AFS server, the partitions containing served files are NOT
     unix filesystems and standard fsck *must* not be used - use the AFS
     version instead.

Subject: 2.02  What is an AFS protection group?

  A named list of users.

  Group names are used in AFS ACLs to identify lists of users with
  particular access permissions.

  In AFS, users can create and maintain their own protection groups.
  This is different to unix where only the system administrator can
  manage /etc/group.

  AFS groups are stored in the protection database on fileserver(s)
  and managed by using the "pts" command.

  An AFS group typically has the format:

      owner-id:group-name

  By default, only the owner of a group can change its members.

  It is possible to have both users and IP addresses as members
  of an AFS group. By using an IP address like this you can specify
  all the users from the host with that IP address.

Subject: 2.03  What are the AFS defined protection groups?

  system:anyuser

      Everyone who has access to an AFS client in any cell that is
      on the same network as your cell.

  system:authuser

      Everyone who has access to an AFS client in any cell that is
      on the same network as your cell *and* has valid tokens for
      your cell (ie has been authenticated in your cell).

  system:administrators

      Users who have privileges to execute some but not all
      system administrator commands.

Subject: 2.04  What is an AFS access control list (ACL)?

  There is an ACL for every directory in AFS. The ACL specifies
  protection at the directory level (not file level) by listing
  permissions of users and/or groups to a directory. There is a
  maximum of 20 entries on an ACL.

  For example:

  An AFS ACL is displayed by using the "fs" command as shown below:

     tweety@toontown $ fs listacl .
     Access list for . is
     Normal rights:
       fac:coords rlidwka
       system:anyuser rl

  This ACL shows that members of the AFS protection group "fac:coords"
  have full access rights to the current directory and "system:anyuser"
  has only read and lookup rights.

  The members of "fac:coords" can be determined by accessing the
  protection group database using the "pts" command as shown below:

     tweety@toontown $ pts membership fac:coords
     Members of fac:coords (id: -1577) are:
       sylvester
       roadrunner
       yosemite.sam

Subject: 2.05  What are the AFS access rights?

  In AFS, there are seven access rights that may be set or not set:

  lookup          l       Permission to examine the ACL and traverse the
                          directory (needed with most other access rights).
                          Permission to look up filenames in a directory.
  read            r       View the contents of files in the directory
  insert          i       Add new files or sub-directories
  write           w       Modify file contents, use "chmod"
  delete          d       Remove file(s) in directory
  lock            k       Permission for programs to "flock" files
                          in the directory
  administer      a       Ability to change the ACL

  There are short-hand forms:

  read            rl      read and lookup
  write           rlidwk  all rights except administer
  all             rlidwka
  none                    removes all rights

Subject: 2.06  What is pagsh?

  A command to get a new shell with a process authentication group (PAG).

  This is normally used if your system does not use the AFS version of login.
  It is used to get a PAG prior to running klog.

  The PAG uniquely identifies the user to the Cache Manager.
  Without a PAG the Cache Manager uses the unix UID to identify a user.

Subject: 2.07  Why use a PAG?

  There are two reasons:

  a) Child processes inherit the PAG and the Kerberos token so they are AFS
     authenticated.

  b) For security: if you don't have a PAG then the Cache Manager identifies
     you by unix UID. Another user with root access to the client could
     su to you and therefore use your token.

Subject: 2.08  How can I tell if I have a PAG?

  You can tell if you have a PAG by typing "groups". A PAG is indicated
  by the appearance of two integers in the list of groups.

  For example:
     sylvester@toontown $ groups
     33536 32533 staff catz

Subject: 2.09  Can I still run cron jobs with AFS?

  Yes, but remember that in order to fully access files in AFS you have
  to be AFS authenticated. If your cron job doesn't klog then it only
  gets system:anyuser access.

  The klog command has a "-pipe" option which will read a password from
  stdin. IF (yes, that's a big if :-) you are prepared to store your
  password in a local (non-AFS) file then you might use the following:

     a) create a "wrapper" script to get a PAG, get your AFS token
        and execute a command:

        #!/usr/afsws/bin/pagsh
        #
        # NAME          afs_wrap_cron
        # AUTHOR        Paul Blackburn <[email protected]>
        # PURPOSE       Run an AFS authenticated cron job.
        #               Get a PAG, get the user's token,
        #               then exec user's command

        CMD=`basename ${0}`

        usage() {
           echo "Usage: ${CMD} [ -principal AFSID ] passwordfile command" >&2
        }

        if [ ${1} = "-principal" ]; then
                PRINCIPAL="${1} ${2}"
                shift 2
        fi

        if [ -z "${1}" ]; then
                echo "${CMD} error: need name of password file" >&2
                usage
                exit 1
        else
                passwordfile=${1}
                shift
        fi

        /usr/afsws/bin/klog ${PRINCIPAL} -pipe < ${passwordfile}

        if [ -z "${1}" ]; then
                echo "${CMD} error: need name of command to run" >&2
                usage
                exit 1
        else
                command_line="$*"
                command=`echo ${command_line} | awk '{print $1}'`

        # Check if we can run the command.
        # If we got this far, it is likely that the command name is correct
        # but there may be a problem in accessing the command file.
        # If there is an error, log it via syslog (logger) rather than ">&2"

                if [ ! -x "${command}" ]; then
                        M="error: unable to execute command ${command}"
                        logger -i -t "${CMD}" "${M}"
                        exit 1
                fi
        fi
        exec ${command_line}

     b) Store your password in a local (non-AFS) file that only you
        have access to (perhaps: /home/$USER/.p).

        Make sure that this file is mode 600 and also be sure that
        you trust whoever has root access on this system and whoever
        has access to backup tapes! Also, don't forget to change this
        file if you change your AFS password.

     c) In your crontab file, run afs_wrap_cron followed by unlog:

        0 6 * * * /usr/local/bin/afs_wrap_cron /home/$USER/.p \
                  $HOME/bin/6AMdaily; /usr/afsws/bin/unlog

  Note that you can still run a cron job without getting a token if
  the task does not need to be AFS authenticated. In this case, you
  may get stderr from the cron job if your .profile is not accessible
  because of the ACL protecting your $HOME. Simply redirect to /dev/null:

        0 7 * * * $sys_anyuser_readable_dir/7AMdaily 2>/dev/null

Subject: 2.10  How much disk space does a 1 byte file occupy in AFS?

  One kilobyte.

  Other filesystems allocate different file block sizes.
  For example, IBM's AIX version 3 journaled file system (JFS)
  uses 4K blocks (exception: 2K for the 160MB disk drive).

  Such blocksize differences lead to variations on the amount of
  disk space required to store files. Copying a directory from AFS
  to AIX JFS would require more space in JFS because of the block
  fragmentation.

  Example:

  a) Create a one byte file in AFS and use "ls -s" to show how many
     kilobytes it occupies:

        ariel@atlantica $ echo z >/afs/dsea/tmp/one_byte_file
        ariel@atlantica $ ls -s /afs/dsea/tmp/one_byte_file
           1 /afs/dsea/tmp/one_byte_file

  b) Create same file in local filesystem (AIX JFS):

        ariel@atlantica $ echo z >/tmp/one_byte_file
        ariel@atlantica $ ls -s /tmp/one_byte_file
           4 /tmp/one_byte_file

Subject: 2.11  Is it possible to specify a user who is external
              to the current AFS cell on an ACL?

  No. You cannot reference a particular user from another AFS cell.

  You can specify an IP address on the ACL; this means any and all
  users from the host with that IP address.

  Another solution to this problem is to give the external user an
  "authentication-only" account in your AFS cell. This means that
  (s)he can klog (but has no home directory) in your cell.

     # Example: AFS administrator creates an authentication-only user
     $ uss add daffy "Daffy Duck" -t /dev/null
     $ kas setpassword daffy -admin admin

  Cross-realm authentication (where co-operating cells are able to
  specify remore users as "[email protected]" on an ACL) is an *unsupported*
  feature of AFS 3.3a. That means that Transarc doesn't promise
  to make it work for you, nor keep it running in future releases.

Subject: 2.12  Are there any problems printing files in /afs?

  The issue of printing in AFS is almost always the same: what do you
  send to the printing daemon?  Do you send it the bytes you want to
  print or do you just send the file name containing those bytes?  If
  you send it a file name, you have to be sure that the printing daemon
  can read it.  Most daemons run with no AFS tokens, so can't access
  directories unless they are open for system:anyuser read access.
  Often, printing commands (lpr, lp, enq) have an option that allows
  for both modes of operation, though the default behavior varies from
  system to system.  If you're interested in making your daemons
  authenticate to AFS, check out the example scripts in AFS-Contrib:

     file:///afs/transarc.com/public/afs-contrib/tools/reauth-example
     ftp://ftp.transarc.com/pub/afs-contrib/tools/reauth-example/

  Another common problem is setuid printing commands.  For instance, the
  "enq" command runs as root, daemon, or some such user.  If you aren't
  using the AFS login and simply issue "klog" to get tokens, those
  tokens are associated with your uid.  When setuid programs run, they
  lose access to your token and often can't read the file name given as
  an argument.  The solution in this case is to use "pagsh" before
  "klog" so that your tokens are transferred to subprocesses
  automatically by group membership.  This works even if the uid
  changes, as for setuid programs.

Subject: 2.13  Can I create a fifo (aka named pipe) in /afs?

  No. AFS does not support "mknod fifofile p".

Subject: 2.14  If an AFS server crashes, do I have to reboot my AFS client?

  No.

  Typically, if an AFS server becomes unavailable, the AFS Cache Manager on
  your AFS client will see you through the outage until the server returns.
  This robustness is dependent on the way your AFS cell has been configured
  including the following factors:

     + On the client side:
        + How big is the cache?
        + Are the files you need already in the cache?

     + On the server side:
        + How many servers? It's best to have a minimum of three.
        + Is the data you are accessing replicated? In AFS, replicas
          are ReadOnly copies.

  With replicated volumes, the AFS Cache Manager knows about all of the
  servers on which the replicas are located. Therefore, when the Cache
  Manager accesses a replicated volume, if the RPC times out, the
  Cache Manager automatically retrys the RPC, using a different file server.

  If necessary, the Cache Manager will attempt to contact all file servers
  on which a replica of the volume resides.

  If you are accessing ReadWrite volumes on a crashed server then you
  will not be able to save changes back to the server until it returns.

  You don't need to reboot, and the Cache Manager activity is "invisible"
  to the user.

Subject: 2.15  Can I use AFS on my diskless workstation?

  Yes. The AFS Cache Manager can be configured to work with either
  a disk based cache or a memory (RAM) based cache. With the latter,
  you can expect file access from the cache with a whizz!

U  http://www.uni-hohenheim.de/~schaefer/afs/info-afs/1306.html

Subject: 2.16  Can I test for AFS tokens from within my program?

  Yes. Some sample code showing how to do this can be found in:

  file:///afs/transarc.com/public/afs-contrib/tools/auth-samples/listtokens.c
  ftp://ftp.transarc.com/pub/afs-contrib/tools/auth-samples/listtokens.c

Subject: 2.17  What's the difference between /afs/cellname and /afs/.cellname?

  AFS has ReadOnly (RO) and ReadWrite (RW) volumes.

  The convention in AFS is to mount the RW volume "root.cell" as
  /afs/.cellname and the RO volume "root.cell.readonly" as /afs/cellname.

  This is so that when you travel down the /afs/.cellname link, AFS will
  always use the RW site of any volumes that have RO clones.

  This allows your administrator to update the RW copy of a volume and
  "vos release $volname" so that it will appear in /afs/cellname.

Subject: 2.18  Can I klog as two users on a machine in the same cell?

  Yes, if you use two different PAGs.

  It's: "One token per PAG per client system."

  From one shell you can only authenticate as a single user of a cell.
  If you open another shell (with another PAG) you can klog as a different
  user of the same cell from the same client.

  You can authenticate into many cells from one client shell.

Subject: 2.19  What are the ~/.__afsXXXX files?

  They are temporary reference files used by the AFS Cache Manager.

  In UNIX filesystems, when you a remove a file that is kept open
  by a process, the file stays around physically while it is no longer
  referenced in any directory (which you will see as a mismatch between
  disk space usage according to df and du).

  Some applications rely on that feature, e.g. they create a temporary file
  and remove it immediatley while keeping the file descriptor open.
  The file then disappears from the filesystem automagically
  when the process terminates or the file descriptor gets closed otherwise.
  Such applications could get into trouble with older versions of AFS,
  where the file could really disappear while it was held open.

  Newer versions of AFS rename such files to .__afsXXXX, thus making sure
  that the data stays around as expected by the application. As soon as
  the file gets closed, the associated .__afsXXXX should disappear.

Subject: 3  AFS administration

Subject: 3.01  Is there a version of xdm available with AFS authentication?

  Yes, xdm can be found in:

     file:///afs/transarc.com/public/afs-contrib/tools/xdm
     ftp://ftp.transarc.com/pub/afs-contrib/tools/xdm/

Subject: 3.02  Is there a version of xlock available with AFS authentication?

  Yes, xlock can be found in:

     file:///afs/transarc.com/public/afs-contrib/tools/xlock
     ftp://ftp.transarc.com/pub/afs-contrib/tools/xlock/

Subject: 3.03  What is /afs/@cell?

  It is a symbolic link pointing at /afs/$your_cell_name.

  NB, @cell is not something that is provided by AFS. You may decide
  it is useful in your cell and wish to create it yourself.

  /afs/@cell is useful because:

     + If you look after more than one AFS cell, you could create
       the link in each cell then set your PATH as:
          PATH=$PATH:/afs/@cell/@sys/local/bin

     + For most cells, it shortens the path names to be typed in
       thus reducing typos and saving time.

  A disadvantage of using this convention is that when you cd into
  /afs/@cell then type "pwd" you see "/afs/@cell" instead of the full name
  of your cell. This may appear confusing if a user wants to tell a user
  in another cell the pathname to a file.

  You could create your own /afs/@cell with the following:

     #/bin/ksh -
     # author: mpb
     [ -L /afs/@cell ] && echo We already have @cell! && exit
     cell=$(cat /usr/vice/etc/ThisCell)
     cd /afs/.${cell} && fs mkm temp root.afs
     cd temp
     ln -s /afs/${cell} @cell
     ln -s /afs/.${cell} .@cell            # .@cell for RW path
     cd /afs/.${cell} && fs rmm temp
     vos release root.afs; fs checkv

U   http://www-archive.stanford.edu/lists/info-afs/hyper95/0298.html

Subject: 3.04  Given that AFS data is location independent, how does
              an AFS client determine which server houses the data
              its user is attempting to access?

   The Volume Location Database (VLDB) is stored on AFS Database
   Servers and is ideally replicated across 3 or more Database Server
   machines.  Replication of the Database ensures high availability
   and load balances the requests for the data.  The VLDB maintains
   information regarding the current physical location of all volume
   data (files and directories) in the cell, including the IP address
   of the FileServer, and the name of the disk partition the data is
   stored on.

   A list of a cell's Database Servers is stored on the local disk of
   each AFS Client machine as: /usr/vice/etc/CellServDB

   The Database Servers also house the Kerberos Authentication
   Database (encrypted user and server passwords), the Protection
   Database (user UID and protection group information) and the
   Backup Database (used by System Administrators to backup AFS file
   data to tape).

Subject: 3.05  Which protocols does AFS use?

  AFS may be thought of as a collection of protocols and software
  processes, nested one on top of the other. The constant interaction
  between and within these levels makes AFS a very sophisticated software
  system.

  At the lowest level is the UDP protocol, which is part of TCP/IP. UDP
  is the connection to the actual network wire. The next protocol level is
  the  remote procedure call (RPC).  In general, RPCs allow the developer
  to build applications using the client/server model, hiding the
  underlying networking mechanisms. AFS uses Rx, an RPC protocol developed
  specifically for AFS during its development phase at Carnegie Mellon
  University.

  Above the RPC is a series of server processes and interfaces that all
  use Rx for communication between machines. Fileserver, volserver,
  upserver, upclient, and bosserver are server processes that export RPC
  interfaces to allow their user interface commands to request actions and
  get information. For example, a bos status <machine name> command will
  examine the bos server process on the indicated file server machine.

  Database servers use ubik, a replicated database mechanism which is
  implemented using RPC. Ubik guarantees that the copies of AFS databases
  of multiple server machines remain consistent. It provides an
  application programming interface (API) for database reads and writes,
  and uses RPCs to keep the database synchronized. The database server
  processes, vlserver, kaserver, and ptserver, reside above ubik. These
  processes export an RPC interface which allows  user commands to control
  their operation.  For instance, the pts command is used to communicate
  with the ptserver, while the command klog  uses the kaserver's RPC
  interface.

  Some application programs are quite complex, and draw on RPC interfaces
  for communication with an assortment of processes. Scout utilizes the
  RPC interface to file server processes to display and monitor the status
  of file servers. The uss command interfaces with  kaserver, ptserver,
  volserver and vlserver to create new user accounts.

  The Cache Manager also exports an RPC interface. This interface is used
  principally by file server machines to break callbacks.  It can also be
  used to obtain Cache Manager status information.  The program cmdebug
  shows the status of a Cache Manager using this interface.

  For additional information, Section 1.5 of the AFS System
  Administrator's Guide and the April 1990 Cache Update contain more
  information on ubik. Udebug information and short descriptions of all
  debugging tools were included in the January 1991 Cache Update. Future
  issues will discuss other debugging tools in more detail.

  [ source: ftp://ftp.transarc.com/pub/afsug/newsletter/apr91 ]
  [ Copyright  1991 Transarc Corporation ]

Subject: 3.06  Are setuid programs executable across AFS cell boundaries?

  By default, the setuid bit is ignored but the program may be run
  (without setuid privilege).

  It is possible to configure an AFS client to honour the setuid bit.
  This is achieved by root running:

     root@toontown # fs setcell -cell $cellname -suid

  (where $cellname is the name of the foreign cell. Use with care!).

  NB: making a program setuid (or setgid) in AFS does *not* mean
  that the program will get AFS permissions of a user or group.
  To become AFS authenticated, you have to klog.  If you are not
  authenticated, AFS treats you as "system:anyuser".

Subject: 3.07  How does AFS maintain consistency on read-write files?

  AFS uses a mechanism called "callback".

  Callback is a promise from the fileserver that the cache version
  of a file/directory is up-to-date. It is established by the fileserver
  with the caching of a file.

  When a file is modified the fileserver breaks the callback.  When the
  user accesses the file again the Cache Manager fetches a new copy
  if the callback has been broken.

  The following paragraphs describe AFS callback mechanism in more detail:

  If I open() fileA and start reading, and you then open() fileA,
  write() a change ***and close() or fsync()*** the file to get your
  changes back to the server - at the time the server accepts and writes
  your changes to the appropriate location on the server disk, the
  server also breaks callbacks to all clients to which it issued a copy
  of fileA.

  So my client receives a message to break the callback on fileA, which
  it dutifully does.  But my application (editor, spreadsheet, whatever
  I'm using to read fileA) is still running, and doesn't really care
  that the callback has been broken.

  When something causes the application to read() more of the file
  the read() system call executes AFS cache manager code via the VFS switch,
  which does check the callback and therefore gets new copies of the data.

  Of course, the application may not re-read data that it has already read,
  but that would also be the case if you were both using the same host.
  So, for both AFS and local files, I may not see your changes.

  Now if I exit the application and start it again, or if the
  application does another open() on the file, then I will see the
  changes you've made.

  This information tends to cause tremendous heartache and discontent
  - but unnecessarily so.  People imagine rampant synchronization problems.
  In practice this rarely happens and in those rare instances, the data in
  question is typically not critical enough to cause real problems or
  crashing and burning of applications.  Since 1985, we've found
  that the synchronization algorithm has been more than adequate in practice
  - but people still like to worry!

  The source of worry is that, if I make changes to a file from my
  workstation, your workstation is not guaranteed to be notified until I
  close or fsync the file, at which point AFS guarantees that your
  workstation will be notified.  This is a significant departure from NFS,
  in which no guarantees are provided.

  Partially because of the worry factor and largely because of Posix,
  this will change in DFS.  DFS synchronization semantics are identical
  to local file system synchronization.

  [ DFS is the Distributed File System which is part of the Distributed ]
  [ Computing Environment (DCE).                                        ]

Subject: 3.08  How can I run daemons with tokens that do not expire?

  It is not a good idea to run with tokens that do not expire because
  this would weaken one of the security features of Kerberos.

  A better approach is to re-authenticate just before the token expires.

  There are two examples of this that have been contributed to
  afs-contrib. The first is "reauth":

  file:///afs/transarc.com/public/afs-contrib/tools/reauth/
  ftp://ftp.transarc.com/pub/afs-contrib/tools/reauth/

  The second is "lat":

  /afs/transarc.com/public/afs-contrib/pointers\
                               /UMich-lat-authenticated-batch-jobs
  ftp://ftp.transarc.com/pub/afs-contrib/pointers
                               /UMich-lat-authenticated-batch-jobs

Subject: 3.09  Can I check my user's passwords for security purposes?

  Yes. Alec Muffett's Crack tool (at version 4.1f) has been converted
  to work on the Transarc kaserver database. This modified Crack
  (AFS Crack) is available via anonymous ftp from:

     ftp://export.acs.cmu.edu/pub/crack.tar.Z

  and is known to work on: pmax_* sun4*_* hp700_* rs_aix* next_*

  It uses the file /usr/afs/db/kaserver.DB0, which is the database on
  the kaserver machine that contains the encrypted passwords. As a bonus,
  AFS Crack is usually two to three orders of magnitude faster than the
  standard Crack since there is no concept of salting in a Kerberos database.

  On a normal UNIX /etc/passwd file, each password can have been encrypted
  around 4096 (2^12) different saltings of the crypt(3) algorithm, so for
  a large number of users it is easy to see that a potentially large
  (up to 4095) number of seperate encryptions of each word checked has
  been avoided.

  Author: Dan Lovinger
  Contact: Derrick J. Brashear <[email protected]>

  Note: AFS Crack does not work for MIT Kerberos Databases.
        The author is willing to give general guidance to someone interested
        in doing the (probably minimal) amount of work to port it to do MIT
        Kerberos. The author does not have access to a MIT Kerberos server
        to do this.

Subject: 3.10  Is there a way to automatically balance disk usage across
              fileservers?

  Yes. There is a tool, balance, which does exactly this.
  It can be retrieved via anonymous ftp from:

     ftp://ftp.andrew.cmu.edu/pub/balance-1.1a.tar.Z

  Actually, it is possible to write arbitrary balancing algorithms
  for this tool. The default set of "agents" provided for the current
  version of balance balance by usage, # of volumes, and activity per week,
  the latter currently requiring a source patch to the AFS volserver.
  Balance is highly configurable.

  Author: Dan Lovinger
  Contact: Derrick Brashear <[email protected]>

Subject: 3.11  Can I shutdown an AFS fileserver without affecting users?

  Yes, this is an example of the flexibility you have in managing AFS.

  Before attempting to shutdown an AFS fileserver you have to make
  some arrangements that any services that were being provided are
  moved to another AFS fileserver:

  1) Move all AFS volumes to another fileserver. (Check you have the space!)
     This can be done "live" while users are actively using files
     in those volumes with no detrimental effects.

  2) Make sure that critical services have been replicated on one
     (or more) other fileserver(s). Such services include:

        kaserver  - Kerberos Authentication server
        vlserver  - Volume Location server
        ptserver  - Protection server
        buserver  - Backup server

     It is simple to test this before the real shutdown by issuing:

        bos shutdown $server $service

     where: $server is the name of the server to be shutdown
       and  $service is one (or all) of: kaserver vlserver ptserver buserver

  Other points to bear in mind:

  + "vos remove" any RO volumes on the server to be shutdown.
    Create corresponding RO volumes on the 2nd fileserver after moving the RW.
    There are two reasons for this:

    1) An RO on the same partition ("cheap replica") requires less space
       than a full-copy RO.

    2) Because AFS always accesses RO volumes in preference to RW,
       traffic will be directed to the RO and therefore quiesce the load
       on the fileserver to be shutdown.

  + If the system to be shutdown has the lowest IP address there may be a
    brief delay in authenticating because of timeout experienced before
    contacting a second kaserver.

Subject: 3.12  How can I set up mail delivery to users with $HOMEs in AFS?

  There are many ways to do this. Here, only two methods are considered:

  Method 1: deliver into local filestore

  This is the simplest to implement. Set up your mail delivery to
  append mail to /var/spool/mail/$USER on one mailserver host.
  The mailserver is an AFS client so users draw their mail out of
  local filestore into their AFS $HOME (eg: inc).

  Note that if you expect your (AFS unauthenticated) mail delivery program
  to be able to process .forward files in AFS $HOMEs then you need to
  add "system:anyuser rl" to each $HOMEs ACL.

  The advantages are:

     + Simple to implement and maintain.
     + No need to authenticate into AFS.

  The drawbacks are:

     - It doesn't scale very well.
     - Users have to login to the mailserver to access their new mail.
     - Probably less secure than having your mailbox in AFS.
     - System administrator has to manage space in /var/spool/mail.

  Method 2: deliver into AFS

  This takes a little more setting up than the first method.

  First, you must have your mail delivery daemon AFS authenticated
  (probably as "postman"). The reauth example in afs-contrib
  shows how a daemon can renew its token. You will also need to setup
  the daemon startup soon after boot time to klog (see the -pipe option).

  Second, you need to set up the ACLs so that "postman" has lookup rights
  down to the user's $HOME and "lik" on $HOME/Mail.

  Advantages:

     + Scales better than first method.
     + Delivers to user's $HOME in AFS giving location independence.
     + Probably more secure than first method.
     + User responsible for space used by mail.

  Disadvantages:

     - More complicated to set up.
     - Need to correctly set ACLs down to $HOME/Mail for every user.
     - Probably need to store postman's password in a file so that
       the mail delivery daemon can klog after boot time.
       This may be OK if the daemon runs on a relatively secure host.

  An example of how to do this for IBM RISC System/6000 is auth-sendmail.
  A beta test version of auth-sendmail can be found in:

  file:///afs/transarc.com/public/afs-contrib/doc/faq/auth-sendmail.tar.Z
  ftp://ftp.transarc.com/pub/afs-contrib/doc/faq/auth-sendmail.tar.Z

Subject: 3.13  Should I replicate a ReadOnly volume on the same partition
              and server as the ReadWrite volume?

  Yes, Absolutely! It improves the robustness of your served volumes.

  If ReadOnly volumes exist (note use of term *exist* rather than
  *are available*), Cache Managers will *never* utilize the ReadWrite
  version of the volume. The only way to access the RW volume is via
  the "dot" path (or by special mounting).

  This means if *all* RO copies are on dead servers, are offline, are
  behind a network partition, etc, then clients will not be able to get
  the data, even if the RW version of the volume is healthy, on a healthy
  server and in a healthy network.

  However, you are *very* strongly encouraged to keep one RO copy of a
  volume on the *same server and partition* as the RW. There are two
  reasons for this:

  1) The RO that is on the same server and partition as the RW is a clone
     (just a copy of the header - not a full copy of each file).
     It therefore is very small, but provides access to the same set of files
     that all other (full copy) ReadOnly volume do.
     Transarc trainers refer to this as the "cheap replica".

  2) To prevent the frustration that occurs when all your ROs are unavailable
     but a perfectly healthy RW was accessible but not used.

     If you keep a "cheap replica", then by definition, if the RW is available,
     one of the RO's is also available, and clients will utilize that site.

Subject: 3.14  Should I start AFS before NFS in /etc/inittab?

  Yes, it is possible to run both AFS and NFS on the same system but
  you should start AFS first.

  In IBM's AIX 3.2, your /etc/inittab would contain:

     rcafs:2:wait:/etc/rc.afs > /dev/console 2>&1 # Start AFS daemons
     rcnfs:2:wait:/etc/rc.nfs > /dev/console 2>&1 # Start NFS daemons

  With AIX, you need to load NFS kernel extensions before the AFS KEs
  in /etc/rc.afs like this:

     #!/bin/sh -
     # example /etc/rc.afs for an AFS fileserver running AIX 3.2
     #
     echo "Installing NFS kernel extensions (for AFS+NFS)"
     /etc/gfsinstall -a /usr/lib/drivers/nfs.ext
     echo "Installing AFS kernel extensions..."
     D=/usr/afs/bin/dkload
     ${D}/cfgexport -a ${D}/export.ext
     ${D}/cfgafs    -a ${D}/afs.ext
     /usr/afs/bin/bosserver &

Subject: 3.15  Will AFS run on a multi-homed fileserver?

  (multi-homed = host has more than one network interface.)

  Yes, it will. However, AFS was designed for hosts with a single IP address.
  There can be problems if you have one host name being resolved to several
  IP addresses.

  Transarc suggest designating unique hostnames for each network interface.
  For example, a host called "spot" has two tokenring and one ethernet
  interfaces: spot-tr0, spot-tr1, spot-en0.
  Then, select which interface will be used for AFS and use that hostname
  in the CellServDB file (eg: spot-tr0).

  You also have to remember to use the AFS interface name with any AFS
  commands that require a server name (eg: vos listvol spot-tr0).

  There is a more detailed discussion of this in the August 1993 issue
  of "Cache Update" (see: ftp://ftp.transarc.com/pub/afsug/newsletter/aug93).

  The simplest way of dealing with this is to make your AFS fileservers
  single-homed (eg only use one network interface).

  At release 3.4 of AFS, it is possible to have multi-homed fileservers
  (but _not_ multi-homed database servers).

Subject: 3.16  Can I replicate my user's home directory AFS volumes?

  No.

  Users with $HOMEs in /afs normally have an AFS ReadWrite volume
  mounted in their home directory.

  You can replicate a RW volume but only as a ReadOnly volume
  and there can only be one instance of a ReadWrite volume.

  In theory, you could have RO copies of a user's RW volume
  on a second server but in practice this won't work for the
  following reasons:

  a) AFS has built-in bias to always access the RO copy of a RW volume.
     So the user would have a ReadOnly $HOME which is not too useful!

  b) Even if a) was not true you would have to arrange frequent
     synchronisation of the RO copy with the RW volume (for example:
     "vos release user.fred; fs checkv") and this would have to be
     done for all such user volumes.

  c) Presumably, the idea of replicating is to recover the $HOME
     in the event of a server crash. Even if a) and b) were not
     problems consider what you might have to do to recover a $HOME:

     1) Create a new RW volume for the user on the second server
        (perhaps named "user.fred.2").

     2) Now, where do you mount it?

        The existing mountpoint cannot be used because it already has
        the ReadOnly copy of the original volume mounted there.

        Let's choose: /afs/MyCell/user/fred.2

     3) Copy data from the RO of the original into the new RW volume
        user.fred.2

     4) Change the user's entry in the password file for the new $HOME:
        /afs/MyCell/user/fred.2

     You would have to attempt steps 1 to 4 for every user who had
     their RW volume on the crashed server. By the time you had done
     all of this, the crashed server would probably have rebooted.

     The bottom line is: you cannot replicate $HOMEs across servers.

Subject: 3.17  Which TCP/IP ports and protocols do I need to enable
              in order to operate AFS through my Internet firewall?

  Assuming you have already taken care of nameserving, you may wish to
  use an Internet timeserver for Network Time Protocol [35] [36]:

     ntp             123/tcp

  A list of NTP servers is available via anonymous FTP from:
     http://www.eecis.udel.edu/~mills/ntp/servers.html

  For further details on NTP see: http://www.eecis.udel.edu/~ntp/

  For a "minimal" AFS service which does not allow inbound or outbound klog:

     fileserver      7000/udp
     cachemanager    7001/udp
     ptserver        7002/udp
     vlserver        7003/udp
     kaserver        7004/udp
     volserver       7005/udp
     reserved        7006/udp
     bosserver       7007/udp

  (Ports in the 7020-7029 range are used by the AFS backup system,
   and won't be needed by external clients performing simple file accesses.)

  Additionally, for "klog" to work through the firewall you need to
  allow inbound and outbound UDP on ports >1024 (probably 1024<port<2048
  would suffice depending on the number of simultaneous klogs).

  See also: http://www-archive.stanford.edu/lists/info-afs/hyper95/0874.html

Subject: 3.18  What is the Andrew Benchmark?

  "It is a script that operates on a collection of files constituting
  an application program. The operations are intended to represent typical
  actions of an average user. The input to the benchmark is a source tree
  of about 70 files. The files total about 200 KB in size. The benchmark
  consists of five distinct phases:

    I MakeDir - Construct a target subtree that is identical to the
                source subtree.
   II Copy    - Copy every file from the source subtree to the target subtree.
  III ScanDir - Traverse the target subtree and examine the status
                of every file in it.
   IV ReadAll - Scan every byte of every file in the target subtree.
    V Make    - Complete and link all files in the target subtree."

  Source:
  file:///afs/transarc.com/public/afs-contrib/doc/benchmark/Andrew.Benchmark.ps
  ftp://ftp.transarc.com/pub/afs-contrib/doc/benchmark/Andrew.Benchmark.ps

Subject: 3.19  Is there a version of HP VUE login with AFS authentication?

  Yes, the availability of this is described in:
     file:///afs/transarc.com/public/afs-contrib/pointers/HP-VUElogin.txt
     ftp://ftp.transarc.com/pub/afs-contrib/pointers/HP-VUElogin.txt

U  If you don't have access to the above, please contact Rajeev Pandey
U  of Hewlett Packard whose email address is <[email protected]>.

Subject: 3.20  How can I list which clients have cached files from a server?

  By using the following script:

  #!/bin/ksh -
  #
  # NAME          afsclients
  # AUTHOR        Rainer Toebbicke  <[email protected]>
  # DATE          June 1994
  # PURPOSE       Display AFS clients which have grabbed files from a server

  if [ $# = 0 ]; then
          echo "Usage: $0 <afs_server 1> ... <afsserver n>"
          exit 1
  fi
  for n; do
          /usr/afsws/etc/rxdebug -servers $n -allconn
  done | grep '^Connection' | \
  while  read x y z ipaddr rest; do echo $ipaddr; done | sort -u |
  while read ipaddr; do
          ipaddr=${ipaddr%%,}
          n="`nslookup $ipaddr`"
          n="${n##*Name: }"
          n="${n%%Address:*}"
          n="${n##*([ ])}"
          n="${n%?}"
          echo "$n ($ipaddr)"
  done

Subject: 3.21  Do Backup volumes require as much space as ReadWrite volumes?

  No.

  The technique used is to create a new volume, where every file in the
  RW copy is pointed to by the new backup volume.  The files don't exist
  in the BK, only in the RW volume.  The backup volume therefore takes up
  very little space.

  If the user now starts modifying data, the old copy must not be destroyed.

  There is a Copy-On-Write bit in the vnode - if the fileserver writes to
  a vnode with the bit on it allocates a new vnode for the data and turns
  off the COW bit. The BK volume hangs onto the old data, and the RW volume
  slowly splits itself away over time.

  The BK volume is re-synchronised with the RW next time a "vos backupsys"
  is run.

  The space needed for the BK volume is directly related to the size
  of all files changed in the RW between runs of "vos backupsys".

Subject: 3.22  Should I run timed on my AFS client?

  No.

  The AFS Cache Manager makes use of NTP [35] [36] to synchronise time
  with your cell's NTP servers.

  Typically, one of your AFS cell's servers synchronises with an
  external NTP server and provides accurate time to your cell.

Subject: 3.23  Why should I keep /usr/vice/etc/CellServDB current?

  On AFS clients, /usr/vice/etc/CellservDB, defines the cells and
  (their servers) that can be accessed via /afs.

  Over time, site details change: servers are added/removed or moved
  onto new network addresses. New sites appear.

  In order to keep up-to-date with such changes, the CellservDB file
  on each AFS client should be kept consistent with some master copy
  (at your site).

  As well as updating CellservDB, your AFS administrator should
  ensure that new cells are mounted in your cell's root.afs volume.

Subject: 3.24  How can I keep /usr/vice/etc/CellServDB current?

  Do a daily copy from a master source and update the AFS kernel sitelist.

  The client CellServDB file must not reside under /afs and is best located
  in local filespace.

  Simply updating a client CellServDB file is not enough.
  You also need to update the AFS kernel sitelist by either:
     a) rebooting the client
  or b) running "fs newcell $cell_name $server_list" for each site in
        the CellServDB file.

  A script to update the AFS kernel sitelist on a running system
  is newCellServDB.

     file:///afs/ece.cmu.edu/usr/awk/Public/newCellServDB
     ftp://ftp.ece.cmu.edu/pub/afs-tools/newCellServDB

  One way to distribute CellServDB is to have a root cron job on each
  AFS client copy the file then run newCellServDB.

  Example:

  #!/bin/ksh -
  #
  # NAME       syncCellServDB
  # PURPOSE    Update local CellServDB file and update AFS kernel sitelist
  # USAGE      run by daily root cron job eg:
  #                    0 3 * * * /usr/local/sbin/syncCellServDB
  #
  # NOTE       "@cell" is a symbolic link to /afs/$this_cell_name

  src=/afs/@cell/service/etc/CellServDB
  dst=/usr/vice/etc/CellServDB
  xec=/usr/local/sbin/newCellServDB
  log=/var/log/syncCellServDB

  if [ -s ${src} ]; then
          if [ ${src} -nt ${dst} ]; then
                  cp $dst ${dst}- && cp $src $dst && $xec 2>&1 >$log
          else
                  echo "master copy no newer: no processing to be done" >$log
          fi
  else
          echo "zero length file: ${src}" >&2
  fi

Subject: 3.25  How can I compute a list of AFS fileservers?

  Here is a Korn shell command to do it:

     stimpy@nick $ vos listvldb -cell $(cat /usr/vice/etc/ThisCell) \
                   | awk '(/server/) {print $2}' | sort -u

Subject: 3.26  How can I set up anonymous FTP login to access /afs?

  The easiest way on a primarily "normal" machine (where you don't want to
  have everything in AFS) is to actually mount root.cell under ~ftp, and then
  symlink /afs to ~ftp/afs or whatever.  It's as simple as changing the
  mountpoint in /usr/vice/etc/cacheinfo and restarting afsd.

  Note that when you do this, anon ftp users can go anywhere system:anyuser
  can (or worse, if you're using IP-based ACLs and the ftp host is PTS groups).
  The only "polite" solution I've arrived at is to have the ftp host
  machine run a minimal CellServDB and police my ACLs tightly.

  Alternatively, you can make ~ftp an AFS volume and just mount whatever you
  need under that - this works well if you can keep everything in AFS, and
  you don't have the same problems with anonymous "escapes" into /afs.

  Unless you need to do authenticating ftp, you are _strongly_ recommended
  using wu-ftpdv2.4 (or better).

Subject: 3.27  Where can I find the Andrew Benchmark?

  file:///afs/transarc.com/public/afs-contrib/doc/faq/ab.tar.Z [156k]
  ftp://ftp.transarc.com/pub/afs-contrib/doc/faq/ab.tar.Z [156k]

  This is a tar archive of file:///afs/cs.cmu.edu/user/satya/ftp/ab/

Subject: 4  Getting more information

Subject: 4.01  Is there an anonymous FTP site with AFS information?

  Yes, it is: ftp.transarc.com

  A brief summary of contents:

  Directory                    Contents

  pub/afsug/newsletter         AFS user group newsletters
  pub/afs-contrib              Contributed tools and documents
  pub/afsps/doc                release notes, SUPPORTED_SYSTEMS.afs.*
  pub/afsug                    AFS user group (see README for detail)
  pub/afsps/progint            AFS programming interface docs

  These directories are also accessible via AFS. For example:
     /afs/transarc.com/public/afs-contrib

  (NB "pub" => "public" when using AFS to access these.)

Subject: 4.02  Which USENET newsgroups discuss AFS?

  alt.filesystems.afs and occasionally in comp.unix.admin.

Subject: 4.03  Where can I get training in AFS?

  Transarc provide user and administrator courses.
  These can be provided at the customer site or at Transarc's offices.

  Transarc's education coordinator may be contacted by:

     telephone: +1 412 338 4363    email: [email protected]

U     http://www.transarc.com

Subject: 4.04  Where can I find AFS resources in World Wide Web (WWW)?

  Here are some I have found (please let me know if you find more):

  a) A collection of AFS information maintained by Derrick Brashear at CMU:

     http://www.cs.cmu.edu/afs/andrew.cmu.edu/usr/shadow/www/afs.html
     (Also accessible in: /afs/andrew.cmu.edu/usr/shadow/www)

  b) AFS Beginners Guide (ALW/NIH):
     http://www.alw.nih.gov/Docs/AFS/AFS_toc.html

  c) NCSA AFS User Guide:
     http://www.ncsa.uiuc.edu/Pubs/UserGuides/AFSGuide/AFSv2.1Book.html

  d) Transarc AFS Product Information:
U     http://www.transarc.com/dfs/public/www/htdocs/.hosts/external/Product/EFS/AFS/afsoverview.html

  e) CERN AFS User's Guide:
     http://wsspinfo.cern.ch/file/doc/afsug.html

  f) MIT SIPB's Inessential AFS
     http://web.mit.edu/afs/sipb.mit.edu/project/doc/afs/html/afs-new.html

  g) Stanford University hypermail archive of [email protected]
     http://www-archive.stanford.edu/lists/info-afs.html

N  h) Linux AFS FAQ:
N     http://www.umlug.umd.edu/linuxafs/

Subject: 4.05  Is there a mailing list for AFS topics?

  Yes, it is [email protected].

  An automated program called Majordomo is now handling the info-afs
  list. To join the mailing list, send a message to:

     [email protected]

  In the body (not the Subject line) of the message, type:

     subscribe info-afs

  For example:

     $ mail -s "subscribe to info-afs" [email protected] <<%
     subscribe info-afs
     %

  To unsubscribe:

     $ mail -s "unsubscribe from info-afs" [email protected] <<%
     unsubscribe info-afs

Subject: 4.06  Where can I find an archive of [email protected]?

     There is a hypermail archive at:
        http://www-archive.stanford.edu/lists/info-afs.html

Subject: 4.07  Where can I find an archive of alt.filesystems.afs?

     file:///afs/ibm.uk/common/archive/alt.filesystems.afs/

  Both the [email protected] and alt.filesystems.afs archives are
  incomplete. If you have material to contribute, please let me know.

Subject: 4.08  Where can I find AFS related GIFs?

U  file:///afs/transarc.com/public/afs-contrib/doc/faq/images/index.html
U  ftp://ftp.transarc.com/pub/afs-contrib/doc/faq/images/index.html

Subject: 4.09  Gibt es eine deutsche AFS Benutzer Gruppe?

  Ja, wenn Sie mitmachen wollen, schicken Sie bitte eine E-Mail an:

     [email protected]

  Ueber diese Adresse werden "subscribe" und "unsubscribe" Requests
  bearbeitet.

Subject: 4.10  Donde puedo encontrar informacion en Espanol sobre AFS?

  Hay algunas notas en Espanol sobre AFS en:
     http://w3.ing.puc.cl/~cet/afs.html

Subject: 5  About the AFS faq

  I started compiling the FAQ after attending an AFS administrators class
  and while waiting for the distribution tape to arrive from Transarc
  (back in July 93). The initial goal was to assist users at my site
  to understand AFS issues.

  The FAQ seemed to be a more widely useful resource so it was made
  generally available.

  I hope you have found the AFS FAQ useful.

  Your criticism or suggestions for improving it are welcome, so please
  don't hesitate to email your views (or just say "hello").

  This compilation is dedicated to my AFS teacher and all those
  who inspire through good humour, enthusiasm, wit and wisdom.
  --
  paul                             http://acm.org/~mpb/homepage.html

Subject: 5.01  How can I get a copy of the AFS faq?

  If you do make a copy, please be aware that this compilation
  changes over time: you will need to do a periodic re-copy to
  keep your copy up-to-date.

  There are two reference sources:

  1) The text only version, available via AFS from:
        /afs/transarc.com/public/afs-contrib/doc/faq/afs.faq

  2) The World Wide Web (HTML) version, available via URL:
U        http://www.angelfire.com/hi/plutonic/afs-faq.html


  There are several other ways to get a copy.

  via AFS: /afs/transarc.com/public/afs-contrib/doc/faq/afs.faq

U  via FTP: ftp://rtfm.mit.edu/pub/usenet/news.answers/afs-faq

U  via WWW: http://www.angelfire.com/hi/plutonic/afs-faq.html

  via USENET news:

           From time to time this faq will be posted to the USENET newsgroups:
              alt.filesystems.afs alt.answers news.answers

  via CD-ROM:

     The AFS faq is now available on CD-ROM "Internet Info" (containing
     17,420 documents including other FAQs, RFCs, IENs, etc) from:

        Walnut Creek CDROM              phone: 1 800 786-9907 (US tollfree)
        4041 Pike Lane, Ste D-www              +1 510 674-0783
        Concord, CA 94250               fax:   +1 510 674-0821
        United States of America        email: [email protected]
                                        WWW:   http://www.cdrom.com/

     The file is in: $cd_mount_point/faqs/alt/filesystems.afs

Subject: 5.02  How can I get my question (and answer) into the AFS faq?

  Comments and contributions are welcome, please send to: [email protected]

  I am looking for reviewers to help me check the material here, please
  let me know if you would like to help.

Subject: 5.03  How can I access the AFS faq via the World Wide Web?

  To access the World Wide Web you either need your own browser
  or have telnet access to WWW servers.

  WWW browsers exist for most machines. Here's a list of some browsers;

     Name     System/requirements           Available from (among others)
     ====     ===================           ==============
     Mosaic   X windows, MS-Windows, Mac    ftp.ncsa.uiuc.edu  /Web
     lynx     vt100                         ftp.wustl.edu /packages/www/lynx

  From your own browser, OPEN or GO to the following document:

U  http://www.angelfire.com/hi/plutonic/afs-faq.html

  It is much better to run your own browser but if this is not possible
  there are several WWW servers accessible via telnet:

  +  telnet info.cern.ch
U     then type: go http://www.angelfire.com/hi/plutonic/afs-faq.html

  +  telnet www.njit.edu                 (login: www)
     then type:
        g
U        http://www.angelfire.com/hi/plutonic/afs-faq.html

  +  telnet  ukanaix.cc.ukans.edu        (login: www, needs vt100)
U     then type: ghttp://www.angelfire.com/hi/plutonic/afs-faq.html

Subject: 6  Bibliography

  If documentation is available via anonymous FTP it is indicated
  by a World Wide Web URL like:

   ftp://athena-dist.mit.edu/pub/kerberos/doc/usenix.PS

   where: athena-dist.mit.edu is the anonymous FTP site and
          pub/kerberos/doc/usenix.PS is the filename

  Similarly, for those who have appropriate access, documents available
  via AFS are shown with the format:

  file:///afs/.....

  [1] John H Howard, Michael L Kazar, Sherri G Menees, David A Nichols,
      M Satyanarayanan, Robert N Sidebotham, Michael J West
      "Scale and Performance in a Distributed File System",
      ACM Transactions on Computer Systems, Vol. 6, No. 1, Feb 1988 pp 51-81.

  [2] Michael L Kazar,
      "Synchronisation and Caching Issues in the Andrew File System",
      USENIX Proceedings, Dallas, TX, Winter 1988

  [3] Alfred Z Spector, Michael L Kazar,
      "Uniting File Systems", UNIX Review, March 1989

  [4] Johna Till Johnson,
      "Distributed File System brings LAN Technology to WANs",
      Data Communications, November 1990, pp 66-67.

  [5] Michael Padovano, PADCOM Associates,
      "AFS widens your horizons in distributed computing",
      Systems Integration, March 1991

  [6] Steve Lammert,
      "The AFS 3.0 Backup System", LISA IV Conference Proceedings,
      Colorado Springs, Colorado, October 1990.

  [7] Michael L Kazar, Bruce W Leverett, Owen T Anderson,
      Vasilis Apostolides, Beth A Bottos, Sailesh Chutani,
      Craig F Everhart, W Anthony Mason, Shu-Tsui Tu, Edward R Zayas,
      "DEcorum File System Architectural Overview",
      USENIX Conference Proceedings, Anaheim, Texas, Summer 1990.

  [8] "AFS Drives DCE Selection", Digital Desktop, Vol 1 No 6 Sept 1990.

  [9] James J Kistler, M Satyanarayanan,
      "Disconnected Operation in the Coda Filesystem",
      CMU School of Computer Science technical report, CMU-CS-91-166
      26th July 1991.

 [10] Puneet Kumar. M Satyanarayanan,
      "Log-based Directory Resolution in the Coda File System",
      CMU School of Computer Science internal document, 2 July 1991.

 [11] Edward R Zayas,
      "Administrative Cells: Proposal for Cooperative Andrew File Systems",
      Information Technology Center internal document,
      Carnegie-Mellon University, 25th June 1987

 [12] Ed Zayas, Craig Everhart,
      "Design and Specification of the Cellular Andrew Environment",
      Information Technology Center, Carnegie-Mellon University,
      CMU-ITC-070, 2 August 1988

 [13] Kazar, Michael L, Information Technology Center,
      Carnegie-Mellon University,
      "Ubik - A library for Managing Ubiquitous Data",
      ITCID, Pittsburgh, PA, 1988

 [14] Kazar, Michael L, Information Technology Center,
      Carnegie-Mellon University,
      "Quorum Completion", ITCID, Pittsburgh, PA, 1988

 [15] SP Miller, BC Neuman, JI Schiller, JH Saltzer,
      "Kerberos Authentication and Authorization System",
      Project Athena technical Plan, Section E.2.1, MIT, December 1987
      ftp://athena-dist.mit.edu/pub/kerberos/doc/techplan.PS
      ftp://athena-dist.mit.edu/pub/kerberos/doc/techplan.txt
      file:///afs/watson.ibm.com/projects/agora/papers/kerberos/techplan.PS

 [16] Bill Bryant,
      "Designing an Authentication System: a Dialogue in Four Scenes",
      Project Athena internal document, MIT, draft of 8th February 1988
      ftp://athena-dist.mit.edu/pub/kerberos/doc/dialogue.PS
      ftp://athena-dist.mit.edu/pub/kerberos/doc/dialogue.mss
      file:///afs/watson.ibm.com/projects/agora/papers/kerberos/dialogue.PS

 [17] Edward R Zayas,
      "AFS-3 Programmer's Reference: Architectural Overview",
      Transarc Corporation, FS-00-D160, September 1991
      ftp://ftp.transarc.com/pub/afsps/doc/progint/archov-doc.ps
      ftp://ftp.transarc.com/pub/afsps/doc/progint/archov-doc.dvi
      file:///afs/transarc.com/public/afsps/doc/progint/archov-doc.ps
      file:///afs/transarc.com/public/afsps/doc/progint/archov-doc.dvi
      file:///afs/watson.ibm.com/projects/agora/papers/afs/archov-doc.ps

 [18] "AFS Programmer's Reference: Authentication Server Interface",
      Transarc Corporation, 12th April 1993
      ftp://ftp.transarc.com/pub/afsps/doc/progint/asrv-ispec.ps
      ftp://ftp.transarc.com/pub/afsps/doc/progint/asrv-ispec.dvi
      file:///afs/transarc.com/public/afsps/doc/progint/asrv-ispec.ps
      file:///afs/transarc.com/public/afsps/doc/progint/asrv-ispec.dvi
      file:///afs/watson.ibm.com/projects/agora/papers/afs/asrv-ispec.ps

 [19] Edward R Zayas,
      "AFS-3 Programmer's Reference: BOS Server Interface",
      Transarc Corporation, FS-00-D161, 28th August 1991
      ftp://ftp.transarc.com/pub/afsps/doc/progint/bsrv-spec.ps
      ftp://ftp.transarc.com/pub/afsps/doc/progint/bsrv-spec.dvi
      file:///afs/transarc.com/public/afsps/doc/progint/bsrv-spec.ps
      file:///afs/transarc.com/public/afsps/doc/progint/bsrv-spec.dvi
      file:///afs/watson.ibm.com/projects/agora/papers/afs/bsrv-spec.ps

 [20] Edward R Zayas,
      "AFS-3 Programmer's Reference: File Server/Cache Manager Interface",
      Transarc Corporation, FS-00-D162, 20th August 1991
      ftp://ftp.transarc.com/pub/afsps/doc/progint/fscm-ispec.ps
      ftp://ftp.transarc.com/pub/afsps/doc/progint/fscm-ispec.dvi
      file:///afs/transarc.com/public/afsps/doc/progint/fscm-ispec.ps
      file:///afs/transarc.com/public/afsps/doc/progint/fscm-ispec.dvi
      file:///afs/watson.ibm.com/projects/agora/papers/afs/fscm-ispec.ps

 [21] Edward R Zayas,
      "AFS-3 Programmer's Reference:
             Specification for the Rx Remote Procedure Call Facility",
      Transarc Corporation, FS-00-D164, 28th August 1991
      ftp://ftp.transarc.com/pub/afsps/doc/progint/rx-spec.ps
      ftp://ftp.transarc.com/pub/afsps/doc/progint/rx-spec.dvi
      file:///afs/transarc.com/public/afsps/doc/progint/rx-spec.ps
      file:///afs/transarc.com/public/afsps/doc/progint/rx-spec.dvi
      file:///afs/watson.ibm.com/projects/agora/papers/afs/rx-spec.ps

 [22] Edward R Zayas,
      "AFS-3 Programmer's Reference:
             Volume Server/Volume Location Server Interface",
      Transarc Corporation, FS-00-D165, 29th August 1991
      ftp://ftp.transarc.com/pub/afsps/doc/progint/vvl-spec.ps
      ftp://ftp.transarc.com/pub/afsps/doc/progint/vvl-spec.dvi
      file:///afs/transarc.com/public/afsps/doc/progint/vvl-spec.ps
      file:///afs/transarc.com/public/afsps/doc/progint/vvl-spec.dvi
      file:///afs/watson.ibm.com/projects/agora/papers/afs/vvl-spec.ps

 [23] "AFS User Guide",
       Transarc Corporation, FS-D200-00.08.3

 [24] "AFS Commands Reference Manual",
       Transarc Corporation, FS-D200-00.11.3

 [25] "AFS Systems Administrators Guide",
       Transarc Corporation, FS-D200-00.10.3

 [26] Steven M. Bellovin, Michael Merritt
      "Limitations of the Kerberos Authentication System",
      Computer Communications Review, October 1990, Vol 20 #5, pp. 119-132
      ftp://research.att.com/dist/internet_security/kerblimit.usenix.ps
      file:///afs/watson.ibm.com/projects/agora/papers/kerberos/limitations.PS

 [27] Jennifer G. Steiner, Clifford Neuman, Jeffrey I. Schiller
      "Kerberos: An Authentication Service for Open Network Systems"
      ftp://athena-dist.mit.edu/pub/kerberos/doc/usenix.PS
      ftp://athena-dist.mit.edu/pub/kerberos/doc/usenix.txt

 [28] Barry Jaspan
      "Kerberos Users' Frequently Asked Questions"
      ftp://rtfm.mit.edu/pub/usenet/news.answers/kerberos-faq/user
      http://www.ov.com/misc/krb-faq.html

 [29] P. Honeyman, L.B. Huston, M.T. Stolarchuk
      "Hijacking AFS"
      ftp://ftp.sage.usenix.org/pub/usenix/winter92/hijacking-afs.ps.Z
      file:///afs/watson.ibm.com/projects/agora/papers/afs/afs_hijacking.ps

 [30] R.N. Sidebotham
      "Rx: Extended Remote Procedure Call"
      Proceedings of the Nationwide File System Workshop
      Information Technology Center, Carnegie Mellon University,
      (August 1988)

 [31] R.N. Sidebotham
      "Volumes: The Andrew File System Data Structuring Primitive"
      Technical Report CMU-ITC-053, Information Technology Center,
      Carnegie Mellon University, (August 1986)

 [32] Cohen, David L. "AFS: NFS on steroids",
      LAN Technology  March 1993 v9 n3 p51(9)
      ftp://ftp.transarc.com/pub/afs-contrib/doc/faq/NFS_on_steroids

 [33] Marybeth Schultz "AFS Troubleshooting Tools"
      Transarc Corporation, January 11 1993, draft document
      ftp://ftp.transarc.com/pub/afsps/doc/trguide/external.afsug.ps

 [34] William Stallings "Kerberos Keeps the Enterprise Secure"
      Data Communications, October 1994, Vol 23 No 14 Page 103

 [35] DL Mills "Internet Time Synchronization: the Network Time Protocol"
      RFC 1129, October 1989
      ftp://nic.ddn.mil/rfc/rfc1129.ps

 [36] DL Mills "Network Time Protocol (Version 3)
         Specification, Implementation and Analysis"
      RFC 1305, March 1992
      ftp://nic.ddn.mil/rfc/rfc1305.tar.Z
      ftp://nic.ddn.mil/rfc/rfc1305.txt

 [37] Craig Everhart "Conventions for Names in the Service Directory
      in the AFS Distributed File System"
      March 1990
      ftp://ftp.transarc.com/pub/afs-contrib/doc/faq/service-spec.ez.ps
      ftp://ftp.transarc.com/pub/afs-contrib/doc/faq/service-spec.ez
      file:///afs/transarc.com/public/afs-contrib/doc/faq/service-spec.ez
      file:///afs/transarc.com/public/afs-contrib/doc/faq/service-spec.ez.ps

 [38] AFS Programmer's Reference Manual
      ftp://ftp.transarc.com/pub/afsps/doc/progref/3.0/
      file:///afs/transarc.com/public/afsps/doc/progref/3.0/

Subject: 7  Change History

  1.82 => 1.83 (Mon Aug 15 15:30:00 1994)
   added: Subject: 0.06  Change History
   added: Subject: 4.06  Where can I find an archive of [email protected]?
   added: Subject: 4.07  Where can I find an archive of alt.filesystems.afs?
   updated: throughout
      changed format of references use WWW URL where appropriate
      checked references point to existing files
   updated: Subject: 0.02  Acknowledgements
      added Michael Fagan and Robert Malick

  1.83 => 1.84 (Mon Aug 15 15:47:22 1994)
   updated: Newsgroups header (fixed typo)

  1.84 => 1.85 (Fri Aug 19 10:30:36 1994)
   updated: Subject: 1.05.e  Single systems image (SSI)
      In example, /usr/afsws/etc/pagsh => /usr/afsws/bin/pagsh for AFS 3.3
   updated: Subject: 1.07  ...Internet AFS filetree?
      added: pi.infn.it mpa-garching.mpg.de pppl.gov meteo.uni-koeln.de
             geo.uni-koeln.de utah.edu
      removed: jrc.flinders.edu.au

  1.85 => 1.86 (Tue Aug 23 17:16:48 1994)
   added: Subject: 3.20  How can I list which clients...
   updated: Subject: 0.02  Acknowledgements
      added Rainer Toebbicke

  1.86 => 1.87 (Mon Oct 10 09:30:00 1994)
   swapped: 0.05 and 0.06
   updated: 1.07  ...Internet AFS filetree?
      added: research.ec.org psu.edu
   added: [32] and [33] to Bibliography
   added: 2.14  If an AFS server crashes,...

  1.87 => 1.88 (Mon Oct 10 12:23:00 1994)
   added: [34] to Bibliography

  1.88 => 1.89 (Thu Oct 13 14:49:21 1994)
   updated: 1.06  Which systems is AFS available for?
      added: Linux

  1.98 => 1.90 (Thu Oct 13 15:07:54 1994)
   updated: 2.14  If an AFS server crashes,...

  1.90 -> 1.91 (Mon Oct 17 08:31:36 1994)
   updated: Table of contents:
      moved: Change History out of Preamble to Subject 7
   updated: 2.14  If an AFS server crashes,...
   updated: 1.06  Which systems is AFS available for?
      added: NetBSD
   added:   2.15  Can I use AFS on my diskless workstation?
   moved: Change History to the end of the file so that a search from
          Table of contents will go straight to the Subject and not
          match on items in the Change History. Also, latest change
          is described at the end of the file.
   updated: all URL refs a la minimal-digest-format FAQ
   updated: 4.04  Where can I find AFS resources in World Wide Web (WWW)?
      added: c) NCSA AFS User Guide
   updated: 5.01  How can I get a copy of the AFS faq?
      added: reference to latest copy via ftp
   added: Anon ftp source reference to first page. (EFAQTOOBIG? :-)

  1.91 => 1.92 (Fri Oct 21 12:03:20 1994)
   added: 2.16  Can I test for AFS tokens from within my program?
   added: 4.08  Where can I find AFS related GIFs?

  1.92 => 1.93 (Wed Dec 14 09:22:32 1994)
   updated: Subject: 0.02  Acknowledgements
      added: David Snearline
   updated: Subject: 0.04  Release Notes
      added ref to changes at end of file
   updated: Subject: 1.01  What is AFS?
      added: Transarc homepage URL
   updated: Subject: 1.05.g  Improved robustness to server crash
   updated: Subject: 1.07  What does "ls /afs" display
      removed: test.alw.nih.gov
      added: afs-math.zib-berlin.de
   updated: Subject: 3.03  How does AFS compare with NFS?
      added: bit about some NFS implementations having extra security
   updated: Subject: 3.12  How can I set up mail delivery...
      added: ref to beta test version of auth-sendmail
   updated: Subject: 3.17  Which TCP/IP ports...
      added: NTP servers URL
   updated: Subject: 4.03  Where can I get training in AFS?
      added: Transarc Training URL
   updated: Subject: 5  About the AFS faq
      added: some history

  1.93 => 1.94 (Thu Jan 05 14:01:04 1995)
   updated: Subject: 5  About the AFS faq
      added: RFF (Request For Feedback)

  1.94 => 1.95 (Wed Jan 11 08:42:50 1995)
   updated: Subject: 0.02  Acknowledgements
      added: Mic Bowman
      added: Mike Prince
   updated: Subject: 1.07
      added: cheme.cmu.edu dia.atd.net desy.de ifh.de
             mathematik-cip.uni-stuttgart.de umr.edu urz.uni-magdeburg.de
      removed: rel-eng.athena.mit.edu nce nce_psc nsf-centers.edu
   updated: Subject: 2.14
      added: bit about ReadWrite volumes
   added: Subject: 3.21  Do Backup volumes...
   updated: Subject: 5.01
      added: new URL for HTML version of AFS FAQ
   updated: Subject: 5.03
      added: new URL for HTML version of AFS FAQ

  1.95 => 1.96 (Wed Jan 11 16:11:28 1995)
   updated: Subject: 5.03
      added: new URL for HTML version of AFS FAQ at Transarc

  1.96 => 1.97 (Tue Jan 17 15:55:37 GMT 1995)
   updated: Subject: 3.12  How can I set up mail delivery...
      now <ftp://grand.central.org/pub/afs-contrib/doc/faq/auth-sendmail.tar.Z>
   updated: Subject: 4.08  Where can I find AFS related GIFs?
      changed to: /afs/transarc.com/public/www/Product/AFS/FAQ/images
   updated: Subject: 5.01 How can I get a copy of the AFS faq?
      changed wording.

  1.97 => 1.98 (Wed Jan 18 13:29:12 GMT 1995)
   updated: Subject: 4.01  Is there an anonymous FTP site with AFS information?
     changed: cache-update to cache.update
   updated: Subject: 4.04  Where can I find AFS resources in World Wide Web?
     added: d) Transarc AFS Product Information
   updated: Subject: 1.07  What does "ls /afs" display
     removed: nce_ctc
     added: wu-wien.ac.at

  1.98 => 1.99 (Tue 28 Feb 14:54:53 1995)
   added: World Wide Web URL reference to first page
   updated: Subject: 1.02  Who supplies AFS?
     added: Transarc homepage URL
   updated: Subject: 1.06  Which systems is AFS available for?
     added: John Kohl and NetBSD mailing list reference
   added: Subject: 1.10  How can I access AFS from my PC?
   updated: Subject: 2.09  Can I still run cron jobs with AFS?
     added: bit about non-authenticated cron jobs
   added: Subject: 2.17  What's the difference between /afs/cellname ...
   updated: Subject: 3.03  How does AFS compare with NFS?
     added: bit about using a RAM based cache and ref to Andrew benchmark
   updated: Subject: 3.17  Which TCP/IP ports and protocols ...
     added: added NTP doc refs
   added: Subject: 3.22  Should I run timed on my AFS client?
   updated: Subject: 4.03  Where can I get training in AFS?
     new (shorter) Transarc URL
   updated: Subject: 4.04  Where can I find AFS resources in (WWW)?
     new (shorter) Transarc URL
   updated: Subject: 5.01  How can I get a copy of the AFS faq?
     new URL http://www.transarc.com/Product/AFS/FAQ/faq.html
   updated: Subject: 5.03  How can I access the AFS faq via the World Wide Web?
     ditto 5.01
   added: [35] [36] to Bibliography

  1.99 => 1.100 (Tue Mar 28 15:31:32 GMT 1995)
   updated: Subject: 4.08  Where can I find AFS related GIFs?
     new (shorter) URL for GIFs
   updated: Subject: 5.01  How can I get a copy of the AFS faq?
     re-worded

  1.100 => 1.101 (Fri Mar 31 07:38:17 GMT 1995)
   updated: Subject: 1.10  How can I access AFS from my PC?
     Added bit about SAMBA.

  1.101 => 1.102 (Tue Apr  4 13:17:54 BST 1995)
   updated: Subject: 0.02  Acknowledgements
     added (New Century Systems) to Rens Troost entry
   updated: Subject: 1.10  How can I access AFS from my PC?
     corrected PCI-Interface URL (pci-faq => pci.faq)

  1.102 => 1.103 (Tue May  9 16:09:32 BST 1995)
   updated: Subject: 5.01  How can I get a copy of the AFS faq?
     Added new CDROM source

  1.103 => 1.104 (Wed May 24 08:43:19 BST 1995)
   updated: Subject: 1.07  What does "ls /afs" display in the Internet AFS
     removed: titech.ac.jp
     added: infn.it iway.org hep.net
   updated: Subject: 3.10  Is there a way to automatically balance disk usage
     added: new version (1.1a) reference
     added: new contact name Derrick Bradshear
   updated: Subject: 4.04  Where can I find AFS resources in WWW?
     added: "CERN AFS User's Guide version 1.0"
   updated: Subject: 5.01  How can I get a copy of the AFS faq?
     added: bit about downloadable HTML version
     added: for CD-ROM source, added both US tollfree and "normal" phone nos.

  1.104 => 1.105 (Thu Jun  8 08:45:56 BST 1995)
   updated: Subject: 3.17  Which TCP/IP ports and protocols
     fixed typo
   updated: Subject: 4.04  Where can I find AFS resources in World Wide Web?
     new URL for Derrick Brashear's AFS reference page
     added: MIT SIPB's Inessential AFS
     added: Getting started with AFS (University of Michigan)

  1.105 => 1.106 (Fri Jul 21 16:56:42 BST 1995)
   throughout: /afs/grand.central.org/pub/afs-contrib changed to
               /afs/transarc.com/public/afs-contrib
   updated: Subject: 0.02  Acknowledgements
     updated: Dan Lovinger now with Microsoft
     added: Anton Knaus (Carnegie Mellon University)
     added: Mike Shaddock (SAS Institute Inc.)
   updated: Subject: 0.04  Release Notes
     changed wording
   updated: Subject: 1.07  What does "ls /afs" display in the Internet AFS
     removed: hepafs1.hep.net inel.gov pub.nsa.hp.com ucop.edu
     added: belwue.uni-tuebingen.de cs.wisc.edu wisc.edu rl.ac.uk asu.edu
            isl.ntt.jp jrc.flinders.edu.au ntt.jp rhic sleeper.nsa.hp.com
            uni-bonn.de
   updated: Subject: 1.08  Why does AFS use Kerberos authentication?
     updated URL and added newsgroup URLs
   updated: Subject: 2.11  Is it possible to specify a user who is external..
     added: example of creating authentication-only account
   moved 3.03 to 1.11: Subject: 1.11  How does AFS compare with NFS?
   replaced old 3.03 by:  Subject: 3.03  What is /afs/@cell?
   updated: Subject: 3.09  Can I check my user's passwords ...
      changed contact and author details
   updated: Subject: 3.10  Is there a way to automatically balance disk usage
      changed contact and author details
   added: Subject: 3.23  Why should I keep /usr/vice/etc/CellServDB current?
   added: Subject: 3.24  How can I keep /usr/vice/etc/CellServDB current?
   added: Subject: 4.09  Bibt es eine deutsche AFS Benutzer Gruppe?

  1.106 => 1.107 (Wed Aug 30 17:20:08 BST 1995)

   added to [37] bibliography

  1.107 => 1.108 (Tue Jan 30 16:59:55 GMT 1996)

   Throughout:
       For the Ohio State U. text to HTML convertor remove "<" ">" from URLs.
       AnonFTP site grand.central.org changed to ftp.transarc.com.
       Fix (most) broken URLs to the restless Web.
       Remove surplus "WWW" strings.
   updated: Subject: 0.02  Acknowledgements
      added: Pat Wilson and Cristian Espinoza
   updated: Subject: 0.05  Quote
      new URL to Othello Act 3 Scene 4 (HTML version)
   updated: Subject: 1.02  Who supplies AFS?
      new URL to Pittsburgh page (HTML version)
   Updated: Subject: 1.07  What does "ls /afs" display in the Internet AFS
      removed: ctd.ornl.gov ntt.jp
      added: bnl.gov dkrz.de fh-heilbronn.de hephy.at zdv.uni-mainz.de
             nrel.gov postech.ac.kr le.caspur.it glue.umd.edu
   updated: Subject: 1.08  Why does AFS use Kerberos authentication?
      new URL to "Using MIT's Kerberos Server with AFS"
   added: Subject: 2.18  Can I klog as two users on a machine in the same cell?
   updated: Subject: 3.07  How does AFS maintain consistency on read-write
      new URLs for DFS and DCE (HTML version)
   updated: Subject: 3.09  Can I check my user's passwords
      Derrick Brashear email [email protected] => [email protected]
   updated: Subject: 3.10  Is there a way to automatically balance disk usage
      Derrick Brashear email [email protected] => [email protected]
   updated: Subject: 3.15  Will AFS run on a multi-homed fileserver?
      add text about AFS 3.4 multi-homed fileservers (not database servers)
   added: Subject: 3.25  How can I compute a list of AFS fileservers?
   added: Subject: 3.26  How can I set up anonymous FTP login to access /afs?
   updated: Subject: 4.01  Is there an anonymous FTP site with AFS information?
      changed ftom grand.central.org to ftp.transarc.com
      added bit about /afs/transarc.com/public/afs-contrib
   updated: Subject: 4.04  Where can I find AFS resources in World Wide Web?
      new URL to Derrick Brashear's AFS reference page
      added http://www-archive.stanford.edu/lists/info-afs.html
   updated: Subject: 4.09 Bibt es eine deutsche AFS Benutzer Gruppe?
      correct "Bibt" to "Gibt"
   added: Subject: 4.10 Donde puedo encontrar informacion en Espanol sobre AFS?

  1.108 => 1.109 (Wed Feb 14 16:55:42 GMT 1996)

   removed: anonymous ftp reference from top (it is in 5.01 anyway)
   updated: Subject: 4.01  Is there an anonymous FTP site with AFS information?
      new location for AFS Programming Interface documentation
   added: [38] to bibliography
   updated: [17] [18] [19] [20] [21] [22] new URLs on ftp.transarc.com
   added: Subject: 3.27  Where can I find the Andrew Benchmark?
   updated: Subject: 1.06  Which systems is AFS available for?
      added URL for OS/2 information

  1.109 => 1.110 (Fri Mar  8 11:02:36 GMT 1996)

   updated: Subject: 0.02  Acknowledgements
      added: Mary Ann DelBusso
   updated: Subject: 2.09  Can I still run cron jobs with AFS?
     new script, afs_wrap_cron, which gets a PAG, token, and execs user command
   updated: Subject: 3.17  Which TCP/IP ports and protocols..firewall..
     added reference to Lyle Seaman's note to [email protected]
   updated: Subject: 3.19  Is there a version of HP VUE login with AFS
     Kevin Eyre now the contact at HP for AFS authenticating HP-VUE
   updated: Subject: 3.25  How can I compute a list of AFS fileservers?
     fixed typo: viva Stimpy!

  1.110 => 1.111 (Fri 29 Aug 16:12:29 1997)

   updated: Subject: 1.06  Which systems is AFS available for?
     mentioned "AFS Client for Windows/NT"
     new mailing list address for Linux and archive URL
   updated: Subject: 4.05  Is there a mailing list for AFS topics?
     new mailing list subscription address for [email protected]

  1.111 => 1.112 (Sat 25 Apr 17:54:00 1998)

   updated: AFS FAQ URL
     was: http://www.transarc.com/Product/AFS/FAQ/faq.html
     now: ftp://ftp.transarc.com/pub/afs-contrib/doc/faq/faq.html
     Changes in Transarc's website made it necessary to move the AFS FAQ URL
   updated: Subject: 0.02  Acknowledgements
      added: Michael Niksch
   updated: Subject: 1.02  Who supplies AFS?
     updated: URL for Pittsburgh (HTML version only)
   Updated: Subject: 1.07  What does "ls /afs" display in the Internet AFS
      removed: cs.arizona.edu dia.atd.net grand.central.org iway.org
               jrc.flinders.edu.au kiewit.dartmouth.edu le.caspur.it
               media-lab.mit.edu pegasus.cranfield.ac.uk
               rus-cip.uni-stuttgart.de ssc.gov stars.com
               stars.reston.unisys.com vfl.paramax.com wisc.edu
      added: bp.ncsu.edu cats.ucsc.edu cert.org cipool.uni-stuttgart.de
             clarkson.edu cs.rose-hulman.edu csv.ica.uni-stuttgart.de
             dapnia.saclay.cea.fr dce.emsl.pnl.gov dce.uni-stuttgart.de
             dis.uniroma1.it enea.it engin.umich.edu eos.ncsu.edu
             federation.atd.net fl.mcs.anl.gov ike.uni-stuttgart.de
             ipp-hgw.mpg.de isk.kth.se jpl.nasa.gov lcp.nrl.navy.mil
             le.infn.it lngs.infn.it msrc.pnl.gov msu.edu phy.bnl.gov
             rz.uni-jena.de spv.uniroma1.it thermo-a.mw.tu-muenchen.de
             tu-bs.de uni-mannheim.de unity.ncsu.edu vn.uniroma3.it
   added: Subject: 2.19  What are the ~/.__afsXXXX files?
   updated: Subject: 3.17 Which TCP/IP ports...firewall
     updated: URL for list of public NTP servers
     added: NTP reference web page

N  1.112 => 1.113 (Thu  9 Jul 19:50:10 1998)
N
N   updated: main header with new URLs
N   updated: Subject: Subject: 0.02  Acknowledgements
N    added: Kelly Chambers (Transarc)
N   updated: Subject: 1.06  Which systems is AFS available for?
N    removed: AFS 3.3 and 3.4 old URL references:
N       ftp://ftp.transarc.com/pub/afsps/doc/SUPPORTED_SYSTEMS.afs.rel33a
N       ftp://ftp.transarc.com/pub/afsps/doc/SUPPORTED_SYSTEMS.afs.rel34
N    added new URL:
N       http://www.transarc.com/Support/afs/relversions/platforms.html
N    added reference to AFS on Linux FAQ at:
N       http://www.umlug.umd.edu/linuxafs/
N   updated: Subject: 1.07  What does "ls /afs" display in the Internet AFS
N      removed: bnl.gov
N      added: dsi.uniroma1.it ovpit.indiana.edu vn.uniroma3.it
N   updated: Subject: 1.10  How can I access AFS from my PC?
N     changed SAMBA URLs:
N       http://samba.anu.edu.au/samba/
N       http://samba.anu.edu.au/samba/docs/faq/sambafaq-1.html#ss1.1
N   updated: Subject: 3.19  Is there a version of HP VUE login with
N                           AFS authentication?
N    removed: reference to:
N       file:///afs/watson.ibm.com/projects/agora/hp/hp-agora/HP-VUElogin/
N    changed HP contact from Kevin Eyre <[email protected]>
N                         to Rajeev Pandey <[email protected]>
N   updated:  Subject: 4.04  Where can I find AFS resources in World Wide Web?
N    changed:
N        http://www.transarc.com/Product/AFS/AFS-Info.html
N     to http://www.transarc.com/dfs/public/www/htdocs/.hosts/external/Product/EFS/AFS/afsoverview.html
N    removed:  http://www.css.itd.umich.edu/docs/tutorials/AFS/
N    added: http://www.umlug.umd.edu/linuxafs/ AFS Linux FAQ
N   updated: Subject: 4.06  Where can I find an archive of info-afs
N    removed: file:///afs/ibm.uk/common/archive/[email protected]/
N   updated: Subject: 4.08  Where can I find AFS related GIFs?
N    changed:
N       file:///afs/transarc.com/public/www/Product/AFS/FAQ/images/
N    to file:///afs/transarc.com/public/afs-contrib/doc/faq/images/
N       http://www.transarc.com/Product/AFS/FAQ/images/index.html
N    to ftp://ftp.transarc.com/pub/afs-contrib/doc/faq/images/index.html
N   updated: Subject: 5.01  How can I get a copy of the AFS faq?
N    changed http URL from:
N        http://www.transarc.com/Product/AFS/FAQ/faq.html
N     to http://www.angelfire.com/hi/plutonic/afs-faq.html
N    removed reference to html.tar.Z and auto email server
N   updated: Subject: 5.03  How can I access the AFS faq via the World Wide Web?
N    changed http URL from:
N        http://www.transarc.com/Product/AFS/FAQ/faq.html
N     to http://www.angelfire.com/hi/plutonic/afs-faq.html

You are viewing proxied material from ftp.icm.edu.pl. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.