jimzah/openafs
1
0
Fork 0
mirror of https://git.openafs.org/openafs.git synced 2025-01-20 07:51:00 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity
81e1d54034
openafs/doc/xml/AdminGuide/c130.html
Chas Williams 52557c982e xml-docbook-documentation-first-pass-20060915 
needs more massaging to make it fit the tree, but, get it here first
2006-09-16 01:13:22 +00:00

1550 lines
53 KiB
HTML
Raw Blame History

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<HTML
><HEAD
><TITLE
>An Overview of AFS Administration</TITLE
><META
NAME="GENERATOR"
CONTENT="Modular DocBook HTML Stylesheet Version 1.7"><LINK
REL="HOME"
TITLE="AFS Administration Guide"
HREF="book1.html"><LINK
REL="UP"
TITLE="Concepts and Configuration Issues"
HREF="p128.html"><LINK
REL="PREVIOUS"
TITLE="Concepts and Configuration Issues"
HREF="p128.html"><LINK
REL="NEXT"
TITLE="Issues in Cell Configuration and Administration"
HREF="c667.html"></HEAD
><BODY
CLASS="chapter"
BGCOLOR="#FFFFFF"
TEXT="#000000"
LINK="#0000FF"
VLINK="#840084"
ALINK="#0000FF"
><DIV
CLASS="NAVHEADER"
><TABLE
SUMMARY="Header navigation table"
WIDTH="100%"
BORDER="0"
CELLPADDING="0"
CELLSPACING="0"
><TR
><TH
COLSPAN="3"
ALIGN="center"
>AFS Administration Guide: Version 3.6</TH
></TR
><TR
><TD
WIDTH="10%"
ALIGN="left"
VALIGN="bottom"
><A
HREF="p128.html"
ACCESSKEY="P"
>Prev</A
></TD
><TD
WIDTH="80%"
ALIGN="center"
VALIGN="bottom"
></TD
><TD
WIDTH="10%"
ALIGN="right"
VALIGN="bottom"
><A
HREF="c667.html"
ACCESSKEY="N"
>Next</A
></TD
></TR
></TABLE
><HR
ALIGN="LEFT"
WIDTH="100%"></DIV
><DIV
CLASS="chapter"
><H1
><A
NAME="HDRWQ5"
></A
>Chapter 1. An Overview of AFS Administration</H1
><P
>This chapter provides a broad overview of the concepts and organization of AFS. It is strongly recommended that anyone
involved in administering an AFS cell read this chapter before beginning to issue commands.</P
><DIV
CLASS="sect1"
><H1
CLASS="sect1"
><A
NAME="HDRWQ6"
>A Broad Overview of AFS</A
></H1
><P
>This section introduces most of the key terms and concepts necessary for a basic understanding of AFS. For a more detailed
discussion, see <A
HREF="c130.html#HDRWQ7"
>More Detailed Discussions of Some Basic Concepts</A
>.</P
><DIV
CLASS="sect2"
><H3
CLASS="sect2"
><A
NAME="AEN137"
>AFS: A Distributed File System</A
></H3
><P
>AFS is a distributed file system that enables users to share and access all of the files stored in a network of
computers as easily as they access the files stored on their local machines. The file system is called distributed for this
exact reason: files can reside on many different machines (be distributed across them), but are available to users on every
machine.</P
></DIV
><DIV
CLASS="sect2"
><H3
CLASS="sect2"
><A
NAME="AEN140"
>Servers and Clients</A
></H3
><P
>In fact, AFS stores files on a subset of the machines in a network, called file server machines. File server machines
provide file storage and delivery service, along with other specialized services, to the other subset of machines in the
network, the client machines. These machines are called clients because they make use of the servers' services while doing
their own work. In a standard AFS configuration, clients provide computational power, access to the files in AFS and other
"general purpose" tools to the users seated at their consoles. There are generally many more client workstations than file
server machines.</P
><P
>AFS file server machines run a number of server processes, so called because each provides a distinct specialized
service: one handles file requests, another tracks file location, a third manages security, and so on. To avoid confusion, AFS
documentation always refers to server machines and server processes, not simply to servers. For a more detailed description of
the server processes, see <A
HREF="c130.html#HDRWQ17"
>AFS Server Processes and the Cache Manager</A
>.</P
></DIV
><DIV
CLASS="sect2"
><H3
CLASS="sect2"
><A
NAME="AEN145"
>Cells</A
></H3
><P
>A cell is an administratively independent site running AFS. As a cell's system administrator, you make many decisions
about configuring and maintaining your cell in the way that best serves its users, without having to consult the
administrators in other cells. For example, you determine how many clients and servers to have, where to put files, and how to
allocate client machines to users.</P
></DIV
><DIV
CLASS="sect2"
><H3
CLASS="sect2"
><A
NAME="AEN148"
>Transparent Access and the Uniform Namespace</A
></H3
><P
>Although your AFS cell is administratively independent, you probably want to organize the local collection of files
(your filespace or tree) so that users from other cells can also access the information in it. AFS enables cells to combine
their local filespaces into a global filespace, and does so in such a way that file access is transparent--users do not need
to know anything about a file's location in order to access it. All they need to know is the pathname of the file, which looks
the same in every cell. Thus every user at every machine sees the collection of files in the same way, meaning that AFS
provides a uniform namespace to its users.</P
></DIV
><DIV
CLASS="sect2"
><H3
CLASS="sect2"
><A
NAME="AEN151"
>Volumes</A
></H3
><P
>AFS groups files into volumes, making it possible to distribute files across many machines and yet maintain a uniform
namespace. A volume is a unit of disk space that functions like a container for a set of related files, keeping them all
together on one partition. Volumes can vary in size, but are (by definition) smaller than a partition.</P
><P
>Volumes are important to system administrators and users for several reasons. Their small size makes them easy to move
from one partition to another, or even between machines. The system administrator can maintain maximum efficiency by moving
volumes to keep the load balanced evenly. In addition, volumes correspond to directories in the filespace--most cells store
the contents of each user home directory in a separate volume. Thus the complete contents of the directory move together when
the volume moves, making it easy for AFS to keep track of where a file is at a certain time. Volume moves are recorded
automatically, so users do not have to keep track of file locations.</P
></DIV
><DIV
CLASS="sect2"
><H3
CLASS="sect2"
><A
NAME="AEN155"
>Efficiency Boosters: Replication and Caching</A
></H3
><P
>AFS incorporates special features on server machines and client machines that help make it efficient and
reliable.</P
><P
>On server machines, AFS enables administrators to replicate commonly-used volumes, such as those containing binaries for
popular programs. Replication means putting an identical read-only copy (sometimes called a clone) of a volume on more than
one file server machine. The failure of one file server machine housing the volume does not interrupt users' work, because the
volume's contents are still available from other machines. Replication also means that one machine does not become
overburdened with requests for files from a popular volume.</P
><P
>On client machines, AFS uses caching to improve efficiency. When a user on a client workstation requests a file, the
Cache Manager on the client sends a request for the data to the File Server process running on the proper file server machine.
The user does not need to know which machine this is; the Cache Manager determines file location automatically. The Cache
Manager receives the file from the File Server process and puts it into the cache, an area of the client machine's local disk
or memory dedicated to temporary file storage. Caching improves efficiency because the client does not need to send a request
across the network every time the user wants the same file. Network traffic is minimized, and subsequent access to the file is
especially fast because the file is stored locally. AFS has a way of ensuring that the cached file stays up-to-date, called a
callback.</P
></DIV
><DIV
CLASS="sect2"
><H3
CLASS="sect2"
><A
NAME="AEN160"
>Security: Mutual Authentication and Access Control Lists</A
></H3
><P
>Even in a cell where file sharing is especially frequent and widespread, it is not desirable that every user have equal
access to every file. One way AFS provides adequate security is by requiring that servers and clients prove their identities
to one another before they exchange information. This procedure, called mutual authentication, requires that both server and
client demonstrate knowledge of a "shared secret" (like a password) known only to the two of them. Mutual authentication
guarantees that servers provide information only to authorized clients and that clients receive information only from
legitimate servers.</P
><P
>Users themselves control another aspect of AFS security, by determining who has access to the directories they own. For
any directory a user owns, he or she can build an access control list (ACL) that grants or denies access to the contents of
the directory. An access control list pairs specific users with specific types of access privileges. There are seven separate
permissions and up to twenty different people or groups of people can appear on an access control list.</P
><P
>For a more detailed description of AFS's mutual authentication procedure, see <A
HREF="c667.html#HDRWQ75"
>A More Detailed
Look at Mutual Authentication</A
>. For further discussion of ACLs, see <A
HREF="c31274.html"
>Managing Access Control
Lists</A
>.</P
></DIV
></DIV
><DIV
CLASS="sect1"
><H1
CLASS="sect1"
><A
NAME="HDRWQ7"
>More Detailed Discussions of Some Basic Concepts</A
></H1
><P
>The previous section offered a brief overview of the many concepts that an AFS system administrator needs to understand.
The following sections examine some important concepts in more detail. Although not all concepts are new to an experienced
administrator, reading this section helps ensure a common understanding of term and concepts.</P
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ8"
>Networks</A
></H2
><P
>A <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>network</I
></SPAN
> is a collection of interconnected computers able to communicate with each other and
transfer information back and forth.</P
><P
>A networked computing environment contrasts with two types of computing environments: <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>mainframe</I
></SPAN
> and
<SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>personal</I
></SPAN
>. <UL
><LI
><P
>A <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>mainframe</I
></SPAN
> computing environment is the most traditional. It uses a single powerful computer
(the mainframe) to do the majority of the work in the system, both file storage and computation. It serves many users,
who access their files and issue commands to the mainframe via terminals, which generally have only enough computing
power to accept input from a keyboard and to display data on the screen.</P
></LI
><LI
><P
>A <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>personal</I
></SPAN
> computing environment is a single small computer that serves one (or, at the most,
a few) users. Like a mainframe computer, the single computer stores all the files and performs all computation. Like a
terminal, the personal computer provides access to the computer through a keyboard and screen.</P
></LI
></UL
></P
><P
>A network can connect computers of any kind, but the typical network running AFS connects high-function personal
workstations. Each workstation has some computing power and local disk space, usually more than a personal computer or
terminal, but less than a mainframe. For more about the classes of machines used in an AFS environment, see <A
HREF="c130.html#HDRWQ10"
>Servers and Clients</A
>.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ9"
>Distributed File Systems</A
></H2
><P
>A <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>file system</I
></SPAN
> is a collection of files and the facilities (programs and commands) that enable users
to access the information in the files. All computing environments have file systems. In a mainframe environment, the file
system consists of all the files on the mainframe's storage disks, whereas in a personal computing environment it consists of
the files on the computer's local disk.</P
><P
>Networked computing environments often use <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>distributed file systems</I
></SPAN
> like AFS. A distributed file
system takes advantage of the interconnected nature of the network by storing files on more than one computer in the network
and making them accessible to all of them. In other words, the responsibility for file storage and delivery is "distributed"
among multiple machines instead of relying on only one. Despite the distribution of responsibility, a distributed file system
like AFS creates the illusion that there is a single filespace.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ10"
>Servers and Clients</A
></H2
><P
>AFS uses a server/client model. In general, a server is a machine, or a process running on a machine, that provides
specialized services to other machines. A client is a machine or process that makes use of a server's specialized service
during the course of its own work, which is often of a more general nature than the server's. The functional distinction
between clients and server is not always strict, however--a server can be considered the client of another server whose
service it is using.</P
><P
>AFS divides the machines on a network into two basic classes, <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>file server machines</I
></SPAN
> and
<SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>client machines</I
></SPAN
>, and assigns different tasks and responsibilities to each.</P
><DIV
CLASS="formalpara"
><P
><B
>File Server Machines: </B
><SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>File server machines</I
></SPAN
> store the files in the distributed file system, and a <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>server
process</I
></SPAN
> running on the file server machine delivers and receives files. AFS file server machines run a number of
<SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>server processes</I
></SPAN
>. Each process has a special function, such as maintaining databases important to AFS
administration, managing security or handling volumes. This modular design enables each server process to specialize in one
area, and thus perform more efficiently. For a description of the function of each AFS server process, see <A
HREF="c130.html#HDRWQ17"
>AFS Server Processes and the Cache Manager</A
>.</P
></DIV
><P
>Not all AFS server machines must run all of the server processes. Some processes run on only a few machines because the
demand for their services is low. Other processes run on only one machine in order to act as a synchronization site. See <A
HREF="c3025.html#HDRWQ90"
>The Four Roles for File Server Machines</A
>.</P
><DIV
CLASS="formalpara"
><P
><B
>Client Machines: </B
>The other class of machines are the <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>client machines</I
></SPAN
>, which generally work directly for users,
providing computational power and other general purpose tools. Clients also provide users with access to the files stored on
the file server machines. Clients do not run any special processes per se, but do use a modified kernel that enables them to
communicate with the AFS server processes running on the file server machines and to cache files. This collection of kernel
modifications is referred to as the Cache Manager; see <A
HREF="c130.html#HDRWQ28"
>The Cache Manager</A
>. There are usually
many more client machines in a cell than file server machines.</P
></DIV
><DIV
CLASS="formalpara"
><P
><B
>Client and Server Configuration: </B
>In the most typical AFS configuration, both file server machines and client machines are high-function workstations
with disk drives. While this configuration is not required, it does have some advantages.</P
></DIV
><P
>There are several advantages to using personal workstations as file server machines. One is that it is easy to expand
the network by adding another file server machine. It is also easy to increase storage space by adding disks to existing
machines. Using workstations rather than more powerful mainframes makes it more economical to use multiple file server
machines rather than one. Multiple file server machines provide an increase in system availability and reliability if popular
files are available on more than one machine.</P
><P
>The advantage of using workstations as clients is that caching on the local disk speeds the delivery of files to
application programs. (For an explanation of caching, see <A
HREF="c130.html#HDRWQ16"
>Caching and Callbacks</A
>.) Diskless
machines can access AFS if they are running NFS(R) and the NFS/AFS Translator, an optional component of the AFS
distribution.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ11"
>Cells</A
></H2
><P
>A <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>cell</I
></SPAN
> is an independently administered site running AFS. In terms of hardware, it consists of a
collection of file server machines and client machines defined as belonging to the cell; a machine can only belong to one cell
at a time. Users also belong to a cell in the sense of having an account in it, but unlike machines can belong to (have an
account in) multiple cells. To say that a cell is administratively independent means that its administrators determine many
details of its configuration without having to consult administrators in other cells or a central authority. For example, a
cell administrator determines how many machines of different types to run, where to put files in the local tree, how to
associate volumes and directories, and how much space to allocate to each user.</P
><P
>The terms <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>local cell</I
></SPAN
> and <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>home cell</I
></SPAN
> are equivalent, and refer to the cell in
which a user has initially authenticated during a session, by logging onto a machine that belongs to that cell. All other
cells are referred to as <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>foreign</I
></SPAN
> from the user's perspective. In other words, throughout a login session,
a user is accessing the filespace through a single Cache Manager--the one on the machine to which he or she initially logged
in--whose cell membership defines the local cell. All other cells are considered foreign during that login session, even if
the user authenticates in additional cells or uses the <SPAN
CLASS="bold"
><B
CLASS="emphasis"
>cd</B
></SPAN
> command to change directories into
their file trees.</P
><P
>It is possible to maintain more than one cell at a single geographical location. For instance, separate departments on a
university campus or in a corporation can choose to administer their own cells. It is also possible to have machines at
geographically distant sites belong to the same cell; only limits on the speed of network communication determine how
practical this is.</P
><P
>Despite their independence, AFS cells generally agree to make their local filespace visible to other AFS cells, so that
users in different cells can share files if they choose. If your cell is to participate in the "global" AFS namespace, it must
comply with a few basic conventions governing how the local filespace is configured and how the addresses of certain file
server machines are advertised to the outside world.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ12"
>The Uniform Namespace and Transparent Access</A
></H2
><P
>One of the features that makes AFS easy to use is that it provides transparent access to the files in a cell's
filespace. Users do not have to know which file server machine stores a file in order to access it; they simply provide the
file's pathname, which AFS automatically translates into a machine location.</P
><P
>In addition to transparent access, AFS also creates a <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>uniform namespace</I
></SPAN
>--a file's pathname is
identical regardless of which client machine the user is working on. The cell's file tree looks the same when viewed from any
client because the cell's file server machines store all the files centrally and present them in an identical manner to all
clients.</P
><P
>To enable the transparent access and the uniform namespace features, the system administrator must follow a few simple
conventions in configuring client machines and file trees. For details, see <A
HREF="c667.html#HDRWQ39"
>Making Other Cells Visible
in Your Cell</A
>.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ13"
>Volumes</A
></H2
><P
>A <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>volume</I
></SPAN
> is a conceptual container for a set of related files that keeps them all together on one
file server machine partition. Volumes can vary in size, but are (by definition) smaller than a partition. Volumes are the
main administrative unit in AFS, and have several characteristics that make administrative tasks easier and help improve
overall system performance. <UL
><LI
><P
>The relatively small size of volumes makes them easy to move from one partition to another, or even between
machines.</P
></LI
><LI
><P
>You can maintain maximum system efficiency by moving volumes to keep the load balanced evenly among the different
machines. If a partition becomes full, the small size of individual volumes makes it easy to find enough room on other
machines for them.</P
></LI
><LI
><P
>Each volume corresponds logically to a directory in the file tree and keeps together, on a single partition, all
the data that makes up the files in the directory. By maintaining (for example) a separate volume for each user's home
directory, you keep all of the user's files together, but separate from those of other users. This is an administrative
convenience that is impossible if the partition is the smallest unit of storage.</P
></LI
><LI
><P
>The directory/volume correspondence also makes transparent file access possible, because it simplifies the process
of file location. All files in a directory reside together in one volume and in order to find a file, a file server
process need only know the name of the file's parent directory, information which is included in the file's pathname.
AFS knows how to translate the directory name into a volume name, and automatically tracks every volume's location, even
when a volume is moved from machine to machine. For more about the directory/volume correspondence, see <A
HREF="c130.html#HDRWQ14"
>Mount Points</A
>.</P
></LI
><LI
><P
>Volumes increase file availability through replication and backup.</P
></LI
><LI
><P
>Replication (placing copies of a volume on more than one file server machine) makes the contents more reliably
available; for details, see <A
HREF="c130.html#HDRWQ15"
>Replication</A
>. Entire sets of volumes can be backed up to tape
and restored to the file system; see <A
HREF="c12776.html"
>Configuring the AFS Backup System</A
> and <A
HREF="c15383.html"
>Backing Up and Restoring AFS Data</A
>. In AFS, backup also refers to recording the state of a
volume at a certain time and then storing it (either on tape or elsewhere in the file system) for recovery in the event
files in it are accidentally deleted or changed. See <A
HREF="c8420.html#HDRWQ201"
>Creating Backup Volumes</A
>.</P
></LI
><LI
><P
>Volumes are the unit of resource management. A space quota associated with each volume sets a limit on the maximum
volume size. See <A
HREF="c8420.html#HDRWQ234"
>Setting and Displaying Volume Quota and Current Size</A
>.</P
></LI
></UL
></P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ14"
>Mount Points</A
></H2
><P
>The previous section discussed how each volume corresponds logically to a directory in the file system: the volume keeps
together on one partition all the data in the files residing in the directory. The directory that corresponds to a volume is
called its <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>root directory</I
></SPAN
>, and the mechanism that associates the directory and volume is called a
<SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>mount point</I
></SPAN
>. A mount point is similar to a symbolic link in the file tree that specifies which volume
contains the files kept in a directory. A mount point is not an actual symbolic link; its internal structure is
different.</P
><DIV
CLASS="note"
><BLOCKQUOTE
CLASS="note"
><P
><B
>Note: </B
>You must not create a symbolic link to a file whose name begins with the number sign (#) or the percent sign (%),
because the Cache Manager interprets such a link as a mount point to a regular or read/write volume, respectively.</P
></BLOCKQUOTE
></DIV
><P
>The use of mount points means that many of the elements in an AFS file tree that look and function just like standard
UNIX file system directories are actually mount points. In form, a mount point is a one-line file that names the volume
containing the data for files in the directory. When the Cache Manager (see <A
HREF="c130.html#HDRWQ28"
>The Cache Manager</A
>)
encounters a mount point--for example, in the course of interpreting a pathname--it looks in the volume named in the mount
point. In the volume the Cache Manager finds an actual UNIX-style directory element--the volume's root directory--that lists
the files contained in the directory/volume. The next element in the pathname appears in that list.</P
><P
>A volume is said to be <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>mounted</I
></SPAN
> at the point in the file tree where there is a mount point pointing
to the volume. A volume's contents are not visible or accessible unless it is mounted.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ15"
>Replication</A
></H2
><P
><SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Replication</I
></SPAN
> refers to making a copy, or <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>clone</I
></SPAN
>, of a source read/write volume
and then placing the copy on one or more additional file server machines in a cell. One benefit of replicating a volume is
that it increases the availability of the contents. If one file server machine housing the volume fails, users can still
access the volume on a different machine. No one machine need become overburdened with requests for a popular file, either,
because the file is available from several machines.</P
><P
>Replication is not necessarily appropriate for cells with limited disk space, nor are all types of volumes equally
suitable for replication (replication is most appropriate for volumes that contain popular files that do not change very
often). For more details, see <A
HREF="c667.html#HDRWQ50"
>When to Replicate Volumes</A
>.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ16"
>Caching and Callbacks</A
></H2
><P
>Just as replication increases system availability, <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>caching</I
></SPAN
> increases the speed and efficiency of
file access in AFS. Each AFS client machine dedicates a portion of its local disk or memory to a cache where it stores data
temporarily. Whenever an application program (such as a text editor) running on a client machine requests data from an AFS
file, the request passes through the Cache Manager. The Cache Manager is a portion of the client machine's kernel that
translates file requests from local application programs into cross-network requests to the <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>File Server
process</I
></SPAN
> running on the file server machine storing the file. When the Cache Manager receives the requested data
from the File Server, it stores it in the cache and then passes it on to the application program.</P
><P
>Caching improves the speed of data delivery to application programs in the following ways:</P
><UL
><LI
><P
>When the application program repeatedly asks for data from the same file, it is already on the local disk. The
application does not have to wait for the Cache Manager to request and receive the data from the File Server.</P
></LI
><LI
><P
>Caching data eliminates the need for repeated request and transfer of the same data, so network traffic is reduced.
Thus, initial requests and other traffic can get through more quickly.</P
></LI
></UL
><P
>While caching provides many advantages, it also creates the problem of maintaining consistency among the many cached
copies of a file and the source version of a file. This problem is solved using a mechanism referred to as a
<SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>callback</I
></SPAN
>.</P
><P
>A callback is a promise by a File Server to a Cache Manager to inform the latter when a change is made to any of the
data delivered by the File Server. Callbacks are used differently based on the type of file delivered by the File Server:
<UL
><LI
><P
>When a File Server delivers a writable copy of a file (from a read/write volume) to the Cache Manager, the File
Server sends along a callback with that file. If the source version of the file is changed by another user, the File
Server breaks the callback associated with the cached version of that file--indicating to the Cache Manager that it
needs to update the cached copy.</P
></LI
><LI
><P
>When a File Server delivers a file from a read-only volume to the Cache Manager, the File Server sends along a
callback associated with the entire volume (so it does not need to send any more callbacks when it delivers additional
files from the volume). Only a single callback is required per accessed read-only volume because files in a read-only
volume can change only when a new version of the complete volume is released. All callbacks associated with the old
version of the volume are broken at release time.</P
></LI
></UL
></P
><P
>The callback mechanism ensures that the Cache Manager always requests the most up-to-date version of a file. However, it
does not ensure that the user necessarily notices the most current version as soon as the Cache Manager has it. That depends
on how often the application program requests additional data from the File System or how often it checks with the Cache
Manager.</P
></DIV
></DIV
><DIV
CLASS="sect1"
><H1
CLASS="sect1"
><A
NAME="HDRWQ17"
>AFS Server Processes and the Cache Manager</A
></H1
><P
>As mentioned in <A
HREF="c130.html#HDRWQ10"
>Servers and Clients</A
>, AFS file server machines run a number of processes,
each with a specialized function. One of the main responsibilities of a system administrator is to make sure that processes are
running correctly as much of the time as possible, using the administrative services that the server processes provide.</P
><P
>The following list briefly describes the function of each server process and the Cache Manager; the following sections
then discuss the important features in more detail.</P
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>File Server</I
></SPAN
>, the most fundamental of the servers, delivers data files from the file server
machine to local workstations as requested, and stores the files again when the user saves any changes to the files.</P
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Basic OverSeer Server (BOS Server)</I
></SPAN
> ensures that the other server processes on its server machine
are running correctly as much of the time as possible, since a server is useful only if it is available. The BOS Server relieves
system administrators of much of the responsibility for overseeing system operations.</P
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Authentication Server</I
></SPAN
> helps ensure that communications on the network are secure. It verifies
user identities at login and provides the facilities through which participants in transactions prove their identities to one
another (mutually authenticate). It maintains the Authentication Database.</P
><P
>The Protection Server helps users control who has access to their files and directories. Users can grant access to several
other users at once by putting them all in a group entry in the Protection Database maintained by the Protection Server.</P
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Volume Server</I
></SPAN
> performs all types of volume manipulation. It helps the administrator move volumes
from one server machine to another to balance the workload among the various machines.</P
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Volume Location Server (VL Server)</I
></SPAN
> maintains the Volume Location Database (VLDB), in which it
records the location of volumes as they move from file server machine to file server machine. This service is the key to
transparent file access for users.</P
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Update Server</I
></SPAN
> distributes new versions of AFS server process software and configuration
information to all file server machines. It is crucial to stable system performance that all server machines run the same
software.</P
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Backup Server</I
></SPAN
> maintains the Backup Database, in which it stores information related to the Backup
System. It enables the administrator to back up data from volumes to tape. The data can then be restored from tape in the event
that it is lost from the file system.</P
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Salvager</I
></SPAN
> is not a server in the sense that others are. It runs only after the File Server or
Volume Server fails; it repairs any inconsistencies caused by the failure. The system administrator can invoke it directly if
necessary.</P
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Network Time Protocol Daemon (NTPD)</I
></SPAN
> is not an AFS server process per se, but plays a vital role
nonetheless. It synchronizes the internal clock on a file server machine with those on other machines. Synchronized clocks are
particularly important for correct functioning of the AFS distributed database technology (known as Ubik); see <A
HREF="c3025.html#HDRWQ103"
>Configuring the Cell for Proper Ubik Operation</A
>. The NTPD is controlled by the <SPAN
CLASS="bold"
><B
CLASS="emphasis"
>runntp</B
></SPAN
> process.</P
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Cache Manager</I
></SPAN
> is the one component in this list that resides on AFS client rather than file
server machines. It not a process per se, but rather a part of the kernel on AFS client machines that communicates with AFS
server processes. Its main responsibilities are to retrieve files for application programs running on the client and to maintain
the files in the cache.</P
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ18"
>The File Server</A
></H2
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>File Server</I
></SPAN
> is the most fundamental of the AFS server processes and runs on each file server
machine. It provides the same services across the network that the UNIX file system provides on the local disk: <UL
><LI
><P
>Delivering programs and data files to client workstations as requested and storing them again when the client
workstation finishes with them.</P
></LI
><LI
><P
>Maintaining the hierarchical directory structure that users create to organize their files.</P
></LI
><LI
><P
>Handling requests for copying, moving, creating, and deleting files and directories.</P
></LI
><LI
><P
>Keeping track of status information about each file and directory (including its size and latest modification
time).</P
></LI
><LI
><P
>Making sure that users are authorized to perform the actions they request on particular files or
directories.</P
></LI
><LI
><P
>Creating symbolic and hard links between files.</P
></LI
><LI
><P
>Granting advisory locks (corresponding to UNIX locks) on request.</P
></LI
></UL
></P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ19"
>The Basic OverSeer Server</A
></H2
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Basic OverSeer Server (BOS Server)</I
></SPAN
> reduces the demands on system administrators by constantly
monitoring the processes running on its file server machine. It can restart failed processes automatically and provides a
convenient interface for administrative tasks.</P
><P
>The BOS Server runs on every file server machine. Its primary function is to minimize system outages. It also</P
><UL
><LI
><P
>Constantly monitors the other server processes (on the local machine) to make sure they are running
correctly.</P
></LI
><LI
><P
>Automatically restarts failed processes, without contacting a human operator. When restarting multiple server
processes simultaneously, the BOS server takes interdependencies into account and initiates restarts in the correct
order.</P
></LI
><LI
><P
>Accepts requests from the system administrator. Common reasons to contact BOS are to verify the status of server
processes on file server machines, install and start new processes, stop processes either temporarily or permanently, and
restart dead processes manually.</P
></LI
><LI
><P
>Helps system administrators to manage system configuration information. The BOS server automates the process of
adding and changing <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>server encryption keys</I
></SPAN
>, which are important in mutual authentication. The BOS
Server also provides a simple interface for modifying two files that contain information about privileged users and
certain special file server machines. For more details about these configuration files, see <A
HREF="c3025.html#HDRWQ85"
>Common
Configuration Files in the /usr/afs/etc Directory</A
>.</P
></LI
></UL
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ20"
>The Authentication Server</A
></H2
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Authentication Server</I
></SPAN
> performs two main functions related to network security: <UL
><LI
><P
>Verifying the identity of users as they log into the system by requiring that they provide a password. The
Authentication Server grants the user a token as proof to AFS server processes that the user has authenticated. For more
on tokens, see <A
HREF="c667.html#HDRWQ76"
>Complex Mutual Authentication</A
>.</P
></LI
><LI
><P
>Providing the means through which server and client processes prove their identities to each other (mutually
authenticate). This helps to create a secure environment in which to send cross-network messages.</P
></LI
></UL
></P
><P
>In fulfilling these duties, the Authentication Server utilizes algorithms and other procedures known as
<SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Kerberos</I
></SPAN
> (which is why many commands used to contact the Authentication Server begin with the letter
<SPAN
CLASS="bold"
><B
CLASS="emphasis"
>k</B
></SPAN
>). This technology was originally developed by the Massachusetts Institute of Technology's
Project Athena.</P
><P
>The Authentication Server also maintains the <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Authentication Database</I
></SPAN
>, in which it stores user
passwords converted into encryption key form as well as the AFS server encryption key. To learn more about the procedures AFS
uses to verify user identity and during mutual authentication, see <A
HREF="c667.html#HDRWQ75"
>A More Detailed Look at Mutual
Authentication</A
>.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ21"
>The Protection Server</A
></H2
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Protection Server</I
></SPAN
> is the key to AFS's refinement of the normal UNIX methods for protecting
files and directories from unauthorized use. The refinements include the following: <UL
><LI
><P
>Defining seven access permissions rather than the standard UNIX file system's three. In conjunction with the UNIX
mode bits associated with each file and directory element, AFS associates an <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>access control list
(ACL)</I
></SPAN
> with each directory. The ACL specifies which users have which of the seven specific permissions for the
directory and all the files it contains. For a definition of AFS's seven access permissions and how users can set them
on access control lists, see <A
HREF="c31274.html"
>Managing Access Control Lists</A
>.</P
></LI
><LI
><P
>Enabling users to grant permissions to numerous individual users--a different combination to each individual if
desired. UNIX protection distinguishes only between three user or groups: the owner of the file, members of a single
specified group, and everyone who can access the local file system.</P
></LI
><LI
><P
>Enabling users to define their own groups of users, recorded in the <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Protection Database</I
></SPAN
>
maintained by the Protection Server. The groups then appear on directories' access control lists as though they were
individuals, which enables the granting of permissions to many users simultaneously.</P
></LI
><LI
><P
>Enabling system administrators to create groups containing client machine IP addresses to permit access when it
originates from the specified client machines. These types of groups are useful when it is necessary to adhere to
machine-based licensing restrictions.</P
></LI
></UL
></P
><P
>The Protection Server's main duty is to help the File Server determine if a user is authorized to access a file in the
requested manner. The Protection Server creates a list of all the groups to which the user belongs. The File Server then
compares this list to the ACL associated with the file's parent directory. A user thus acquires access both as an individual
and as a member of any groups.</P
><P
>The Protection Server also maps usernames (the name typed at the login prompt) to <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>AFS user ID</I
></SPAN
>
numbers (<SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>AFS UIDs</I
></SPAN
>). These UIDs are functionally equivalent to UNIX UIDs, but operate in the domain of AFS
rather than in the UNIX file system on a machine's local disk. This conversion service is essential because the tokens that
the Authentication Server grants to authenticated users are stamped with usernames (to comply with Kerberos standards). The
AFS server processes identify users by AFS UID, not by username. Before they can understand whom the token represents, they
need the Protection Server to translate the username into an AFS UID. For further discussion of tokens, see <A
HREF="c667.html#HDRWQ75"
>A More Detailed Look at Mutual Authentication</A
>.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ22"
>The Volume Server</A
></H2
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Volume Server</I
></SPAN
> provides the interface through which you create, delete, move, and replicate
volumes, as well as prepare them for archiving to tape or other media (backing up). <A
HREF="c130.html#HDRWQ13"
>Volumes</A
>
explained the advantages gained by storing files in volumes. Creating and deleting volumes are necessary when adding and
removing users from the system; volume moves are done for load balancing; and replication enables volume placement on multiple
file server machines (for more on replication, see <A
HREF="c130.html#HDRWQ15"
>Replication</A
>).</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ23"
>The Volume Location (VL) Server</A
></H2
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>VL Server</I
></SPAN
> maintains a complete list of volume locations in the <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Volume Location
Database (VLDB)</I
></SPAN
>. When the Cache Manager (see <A
HREF="c130.html#HDRWQ28"
>The Cache Manager</A
>) begins to fill a
file request from an application program, it first contacts the VL Server in order to learn which file server machine
currently houses the volume containing the file. The Cache Manager then requests the file from the File Server process running
on that file server machine.</P
><P
>The VLDB and VL Server make it possible for AFS to take advantage of the increased system availability gained by using
multiple file server machines, because the Cache Manager knows where to find a particular file. Indeed, in a certain sense the
VL Server is the keystone of the entire file system--when the information in the VLDB is inaccessible, the Cache Manager
cannot retrieve files, even if the File Server processes are working properly. A list of the information stored in the VLDB
about each volume is provided in <A
HREF="c8420.html#HDRWQ180"
>Volume Information in the VLDB</A
>.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ24"
>The Update Server</A
></H2
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Update Server</I
></SPAN
> helps guarantee that all file server machines are running the same version of a
server process. System performance can be inconsistent if some machines are running one version of the BOS Server (for
example) and other machines were running another version.</P
><P
>To ensure that all machines run the same version of a process, install new software on a single file server machine of
each system type, called the <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>binary distribution machine</I
></SPAN
> for that type. The binary distribution machine
runs the server portion of the Update Server, whereas all the other machines of that type run the client portion of the Update
Server. The client portions check frequently with the <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>server portion</I
></SPAN
> to see if they are running the right
version of every process; if not, the <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>client portion</I
></SPAN
> retrieves the right version from the binary
distribution machine and installs it locally. The system administrator does not need to remember to install new software
individually on all the file server machines: the Update Server does it automatically. For more on binary distribution
machines, see <A
HREF="c3025.html#HDRWQ93"
>Binary Distribution Machines</A
>.</P
><P
>In cells that run the United States edition of AFS, the Update Server also distributes configuration files that all file
server machines need to store on their local disks (for a description of the contents and purpose of these files, see <A
HREF="c3025.html#HDRWQ85"
>Common Configuration Files in the /usr/afs/etc Directory</A
>). As with server process software, the need
for consistent system performance demands that all the machines have the same version of these files. With the United States
edition, the system administrator needs to make changes to these files on one machine only, the cell's <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>system
control machine</I
></SPAN
>, which runs a server portion of the Update Server. All other machines in the cell run a client
portion that accesses the correct versions of these configuration files from the system control machine. Cells running the
international edition of AFS do not use a system control machine to distribute configuration files. For more information, see
<A
HREF="c3025.html#HDRWQ94"
>The System Control Machine</A
>.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ25"
>The Backup Server</A
></H2
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Backup Server</I
></SPAN
> maintains the information in the <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Backup Database</I
></SPAN
>. The Backup
Server and the Backup Database enable administrators to back up data from AFS volumes to tape and restore it from tape to the
file system if necessary. The server and database together are referred to as the Backup System.</P
><P
>Administrators initially configure the Backup System by defining sets of volumes to be dumped together and the schedule
by which the sets are to be dumped. They also install the system's tape drives and define the drives' <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Tape
Coordinators</I
></SPAN
>, which are the processes that control the tape drives.</P
><P
>Once the Backup System is configured, user and system data can be dumped from volumes to tape. In the event that data is
ever lost from the system (for example, if a system or disk failure causes data to be lost), administrators can restore the
data from tape. If tapes are periodically archived, or saved, data can also be restored to its state at a specific time.
Additionally, because Backup System data is difficult to reproduce, the Backup Database itself can be backed up to tape and
restored if it ever becomes corrupted. For more information on configuring and using the Backup System, see <A
HREF="c12776.html"
>Configuring the AFS Backup System</A
> and <A
HREF="c15383.html"
>Backing Up and Restoring AFS
Data</A
>.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ26"
>The Salvager</A
></H2
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Salvager</I
></SPAN
> differs from other AFS Servers in that it runs only at selected times. The BOS Server
invokes the Salvager when the File Server, Volume Server, or both fail. The Salvager attempts to repair disk corruption that
can result from a failure.</P
><P
>As a system administrator, you can also invoke the Salvager as necessary, even if the File Server or Volume Server has
not failed. See <A
HREF="c8420.html#HDRWQ232"
>Salvaging Volumes</A
>.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ27"
>The Network Time Protocol Daemon</A
></H2
><P
>The <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Network Time Protocol Daemon (NTPD)</I
></SPAN
> is not an AFS server process per se, but plays an important
role. It helps guarantee that all of the file server machines agree on the time. The NTPD on one file server machine acts as a
synchronization site, generally learning the correct time from a source outside the cell. The NTPDs on the other file server
machines refer to the synchronization site to set the internal clocks on their machines.</P
><P
>Keeping clocks synchronized is particularly important to the correct operation of AFS's distributed database technology,
which coordinates the copies of the Authentication, Backup, Protection, and Volume Location Databases; see <A
HREF="c667.html#HDRWQ52"
>Replicating the AFS Administrative Databases</A
>. Client machines also refer to these clocks for the
correct time; therefore, it is less confusing if all file server machines have the same time. For more technical detail about
the NTPD, see <A
HREF="c6449.html#HDRWQ151"
>The runntp Process</A
>.</P
></DIV
><DIV
CLASS="sect2"
><H2
CLASS="sect2"
><A
NAME="HDRWQ28"
>The Cache Manager</A
></H2
><P
>As already mentioned in <A
HREF="c130.html#HDRWQ16"
>Caching and Callbacks</A
>, the <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>Cache Manager</I
></SPAN
> is
the one component in this section that resides on client machines rather than on file server machines. It is not technically a
stand-alone process, but rather a set of extensions or modifications in the client machine's kernel that enable communication
with the server processes running on server machines. Its main duty is to translate file requests (made by application
programs on client machines) into <SPAN
CLASS="emphasis"
><I
CLASS="emphasis"
>remote procedure calls (RPCs)</I
></SPAN
> to the File Server. (The Cache Manager
first contacts the VL Server to find out which File Server currently houses the volume that contains a requested file, as
mentioned in <A
HREF="c130.html#HDRWQ23"
>The Volume Location (VL) Server</A
>). When the Cache Manager receives the requested
file, it caches it before passing data on to the application program.</P
><P
>The Cache Manager also tracks the state of files in its cache compared to the version at the File Server by storing the
callbacks sent by the File Server. When the File Server breaks a callback, indicating that a file or volume changed, the Cache
Manager requests a copy of the new version before providing more data to application programs.</P
></DIV
></DIV
></DIV
><DIV
CLASS="NAVFOOTER"
><HR
ALIGN="LEFT"
WIDTH="100%"><TABLE
SUMMARY="Footer navigation table"
WIDTH="100%"
BORDER="0"
CELLPADDING="0"
CELLSPACING="0"
><TR
><TD
WIDTH="33%"
ALIGN="left"
VALIGN="top"
><A
HREF="p128.html"
ACCESSKEY="P"
>Prev</A
></TD
><TD
WIDTH="34%"
ALIGN="center"
VALIGN="top"
><A
HREF="book1.html"
ACCESSKEY="H"
>Home</A
></TD
><TD
WIDTH="33%"
ALIGN="right"
VALIGN="top"
><A
HREF="c667.html"
ACCESSKEY="N"
>Next</A
></TD
></TR
><TR
><TD
WIDTH="33%"
ALIGN="left"
VALIGN="top"
>Concepts and Configuration Issues</TD
><TD
WIDTH="34%"
ALIGN="center"
VALIGN="top"
><A
HREF="p128.html"
ACCESSKEY="U"
>Up</A
></TD
><TD
WIDTH="33%"
ALIGN="right"
VALIGN="top"
>Issues in Cell Configuration and Administration</TD
></TR
></TABLE
></DIV
></BODY
></HTML
>
Reference in New Issue View Git Blame Copy Permalink