329 lines
12 KiB
Plaintext
329 lines
12 KiB
Plaintext
/*
|
|
* Copyright (c) 1992, 1993, 1994 by the University of Southern California
|
|
*
|
|
* For copying and distribution information, please see the file
|
|
* <usc-license.h>
|
|
*
|
|
* Written by bcn 1989 modified 1989-1992
|
|
* Modified by bcn 1/19/93 to eliminate server specific definitions
|
|
* Modified by swa 5/2/94 to add section 3 and reorder things to be less
|
|
* intimidating to neophytes.
|
|
*/
|
|
|
|
#include <usc-license.h>
|
|
|
|
/*
|
|
* psite.h - site selected parameters for Prospero
|
|
*
|
|
* This file contains local definitions for the Prospero utilities.
|
|
* It is expected that it will be different on different systems.
|
|
*/
|
|
|
|
/*
|
|
* Section 1: Installation specifc definitions for the Prospero Applications
|
|
*
|
|
* The definitions in the following section provide information about
|
|
* the proper configuration of Prospero clients on your system.
|
|
* They may change from system to system.
|
|
*/
|
|
|
|
/*
|
|
* If P_BINARIES is defined and not empty, it defines the directory
|
|
* in which the Prospero clients will look for executables. If set in
|
|
* the users environment, P_BINARIES will override the definition
|
|
* here. If neither are defined, or if the definition is the empty
|
|
* string, then the clients will use the user's search path to find
|
|
* the binaries.
|
|
*
|
|
* The only executable program that the prospero clients and compatability
|
|
* library call currently is VCACHE.
|
|
*/
|
|
#define P_BINARIES "/usr/pfs/bin"
|
|
|
|
/*
|
|
* If P_PATH, if defined, is a directory that will be added to the
|
|
* user's search path by the Prospero initialization script
|
|
* (vfsetup.source for CSH users, vfsetup.profil for SH users).
|
|
* It only needs to be defined if the Prospero binaries will not already be
|
|
* in the user's search path. Multiple directories separated by spaces may
|
|
* be specified.
|
|
*/
|
|
#define P_PATH P_BINARIES
|
|
|
|
|
|
/*
|
|
* P_AFS is the name of the directory under which AFS files are found.
|
|
* If should be set only if your system supports the Andrew File System.
|
|
*
|
|
* #define P_AFS "/afs"
|
|
*/
|
|
|
|
/*
|
|
* P_KERBEROS, if defined, means that the clients will transmit Kerberos
|
|
* (version 5) authentication information and that the server will understand
|
|
* such information. This should be #defined if your site uses Kerberos
|
|
* authentication; otherwise, you won't have a use for it. If you #define
|
|
* this, then you also need to configure the top-level Makefile with the
|
|
* location of the Kerberos libraries and include directory.
|
|
*
|
|
* Note that most Kerberos sites out there are still using the older version 4,
|
|
* which is not compatible with the new version 5.
|
|
*
|
|
* #define P_KERBEROS
|
|
*/
|
|
|
|
|
|
/*
|
|
* P_P_PASSWORD, if defined, means that the clients will transmit password
|
|
* authentication information.
|
|
*
|
|
* #define P_P_PASSWORD
|
|
*/
|
|
|
|
/*
|
|
* P_ACCOUNT, if defined, means that the clients will transmit
|
|
* accounting information.
|
|
*
|
|
* #define P_ACCOUNT
|
|
*/
|
|
|
|
|
|
|
|
|
|
/*
|
|
* Section 2: Site Definitions for the Prospero Applications
|
|
*
|
|
* The definitions in the following section define the Prospero site.
|
|
* They will not necessarily change from system to system. These definitions
|
|
* should only be changed if you are setting up your own prospero site.
|
|
* Setting up a site is different than installing the software on a new
|
|
* system. Unless your site supports storage of new virtual systems,
|
|
* do not change these definitions.
|
|
*/
|
|
|
|
/* Section 2a: Site definitions we expect you might need to change
|
|
* for your site.
|
|
*/
|
|
|
|
/*
|
|
* P_SITE_HOST and DIRECTORY define the system level name of the
|
|
* virtual directory under which site data and meta-data are stored.
|
|
* It will vary from site to site.
|
|
*/
|
|
#define P_SITE_HOST "PROSPERO.ISI.EDU"
|
|
#define P_SITE_DIRECTORY "/pfs/pfsdat/guest"
|
|
|
|
/*
|
|
* P_SITE_MASTER_VSLIST is the name of the virtual directory (relative
|
|
* to P_SITE) under which reference to local virtual systems can be
|
|
* found. When setting up a new site, choose a one component name
|
|
* that seems appropriate.
|
|
*/
|
|
#define P_SITE_MASTER_VSLIST "vs_root_isi-guest"
|
|
|
|
/*
|
|
* Section 2b: Site Definitions we expect you to not change except in unusual
|
|
* circumstances.
|
|
*/
|
|
|
|
/*
|
|
* P_GLOBAL_VS_ROOT* is a reference to the global root for the
|
|
* underlying ugly-names of virtual systems. These definitions should not
|
|
* be changed unless your site is not connected to the Internet.
|
|
*/
|
|
#define P_GLOBAL_VS_ROOT_HOST "PROSPERO.ISI.EDU"
|
|
#define P_GLOBAL_VS_ROOT_FILE "/pfs/pfsdat/vs_root_g"
|
|
|
|
/*
|
|
* Section 2c: Site definitions you should never have a need to change
|
|
* (they can be reconfigured because we did not want
|
|
* Prospero to have any particular directory names hardcoded
|
|
* into it, and it will not hurt Prospero in any way if you
|
|
* change them, but we think it is unlikely you would want to..)
|
|
*/
|
|
|
|
/*
|
|
* P_MASTER_VS is the name (relative to MASTER_VSLIST) of the virtual
|
|
* system to be used when creating new virtual systems with newvs.
|
|
* There should be no need to change these definitions.
|
|
*/
|
|
#define P_MASTER_VS "master"
|
|
#define P_PROTOTYPE_VS "prototype"
|
|
|
|
/*
|
|
* P_SITE_STORAGE is the name (relative to master) of a directory
|
|
* with references to the storage areas of the individual systems that
|
|
* compose a site. There should be no need to change this
|
|
* definitions.
|
|
*/
|
|
#define P_SITE_STORAGE "pfs_storage"
|
|
|
|
/*
|
|
* P_VS_STORAGE is the name (relative to the directories named in
|
|
* P_SITE_STORAGE) of the directory under which new virtual systems
|
|
* may be stored. There should be no need to change this definition.
|
|
*/
|
|
#define P_VS_STORAGE "local_vsystems"
|
|
|
|
/*
|
|
* Section 3: Special-purpose configuration options for the Prospero
|
|
* Applications
|
|
*
|
|
*
|
|
* The configuration options in this section are used only to meet special
|
|
* needs. They have been added to meet the special needs of particular
|
|
* Prospero commercial and research users. They have general utility, but they
|
|
* are more in the nature of optimizations and tuning. Prospero will function
|
|
* just fine if you leave them alone, although not as efficiently in some
|
|
* applications.
|
|
*
|
|
* Unless you are an experienced Prospero user, you do not want to
|
|
* alter any of these definitions.
|
|
*/
|
|
|
|
/*
|
|
* If NFS is supported P_NFS can defined. If set, the clients will attempt
|
|
* to use NFS to retrieve files from remote systems.
|
|
*
|
|
* *WARNING* *WARNING* *WARNING* *WARNING* *WARNING* *WARNING* *WARNING*
|
|
*
|
|
* If you will be using this option, you need to modify the
|
|
* function pmap_nfs() in the file lib/psrv/pmap_nfs.c to do proper NFS
|
|
* retrievals from the server. You also will need to modify the function
|
|
* check_nfs() in the file lib/psrv/check_nfs.c to make sure that your
|
|
* server publishes NFS data only for exportable filesystems, and that it
|
|
* publishes the data with the correct NFS filesystem names. The version of
|
|
* lib/psrv/check_nfs.c we provide is only a stub, and does not perform
|
|
* this function appropriately for most environments.
|
|
*
|
|
* This option is growing obsolescent with the existence of automounters.
|
|
* It's still useful for using sites such as WU-ARCHIVE, which allow one to
|
|
* anonymously mount NFS partitions, and you are welcome to experiment with it
|
|
* for those purposes.
|
|
*
|
|
* *WARNING* *WARNING* *WARNING* *WARNING* *WARNING* *WARNING* *WARNING*
|
|
*
|
|
* Until you customize lib/psrv/check_nfs.c and lib/psrv/pmap_nfs.c, do NOT
|
|
* #define P_NFS. If your clients run an NFS automounter,
|
|
* you can accomplish the same result by defining SHARED_PREFIXES in the
|
|
* pserver.h file.
|
|
*
|
|
*
|
|
* #define P_NFS
|
|
*/
|
|
|
|
/*
|
|
* P_CACHE_* define locations for files associated with the caching on
|
|
* the client.
|
|
*
|
|
* P_CACHE_TMP is for temporary files during transfer,
|
|
* P_CACHE_VC is for the cache itself
|
|
* P_CACHE_P is for personal copies of files to ensure they don't change
|
|
*
|
|
* Allthough these can be symbolic links, they MUST all be in the same
|
|
* filesystem since hard-links are used to move them around
|
|
*
|
|
* They must also match the definitions at the top of Makefile
|
|
*
|
|
* Note that each of these directories needs to be writable by whatever ids
|
|
* are running Prospero clients. Directories under P_CACHE_P can be restricted
|
|
* to specific users, and if "vcache" was to run setuid "pfs" then
|
|
* P_CACHE_VC could be restricted to that user, although code would
|
|
* then be needed in vcache to chown files to the particular user -- and
|
|
* that won't be possible on some (e.g. Sun) unixes.
|
|
* (swa: I don't understand why mitra made the comment above. This can be
|
|
* accomplished by setting the sticky bit on directories.)
|
|
*
|
|
* This caching code was provided by mitra@path.net and is still experimental.
|
|
* It is not a supported part of this release.
|
|
*
|
|
* It lacks a number of facilities that willl have to be present before this
|
|
* becomes a supported part of the release. Send email to
|
|
* info-prospero@ISI.EDU if you want to know what changes need to be made.
|
|
*
|
|
* The strategy for expiring items from the cache is in lib/pfs/pmap_cache.c.
|
|
*
|
|
* #define P_CACHE_ENABLED
|
|
*/
|
|
|
|
/* These definitions will have no effect unless P_CACHE_ENABLED is defined. */
|
|
#ifdef P_CACHE_ENABLED
|
|
#define P_CACHE_P "/nfs/u5/divirs/vcache/p"
|
|
#define P_CACHE_VC "/nfs/u5/divirs/vcache/vc"
|
|
#define P_CACHE_TMP "/nfs/u5/divirs/vcache/tmp"
|
|
#endif
|
|
|
|
|
|
/*
|
|
* This option is not currently supported in the release, because it needs
|
|
* cleanup; notes available from SWA on what steps need to be taken to clean
|
|
* it up to be of production quality.
|
|
*
|
|
* It lacks a number of facilities that will have to be present before this
|
|
* becomes a supported part of the release. Send email to
|
|
* info-prospero@ISI.EDU if you want to know what changes need to be made.
|
|
*
|
|
* You can have the ACCESS-METHOD interpreting code that does retrievals
|
|
* linked directly with your client executables. This significantly bloats
|
|
* executable size, but gains you more speed on the retrievals (as long as you
|
|
* manage to avoid thrashing.) This optimization is useful in circumstances
|
|
* where the cost of slower response is greater than the cost of buying some
|
|
* more RAM for your machine.
|
|
*
|
|
* N.B.: if your machine is paging or swapping
|
|
* under these usage conditions, this optimization will cost
|
|
* you time rather than saving it, unless you are able to configure
|
|
* libpvcache on your system to be a shared library where only one copy of the
|
|
* library exists in RAM for all of the clients.
|
|
*
|
|
* This is a traditional space/time tradeoff.
|
|
* To set the tradeoff in favor of reducing time and increasing space,
|
|
* define INCREASE_CLIENT_EXECUTABLE_SIZE_DO_NOT_EXEC_VCACHE. By default,
|
|
* a separate executable named 'vcache' is run to perform some types of
|
|
* retrievals.
|
|
*
|
|
* If you define this, then also edit VCACHE_LIBS in the top-level Prospero
|
|
* Makefile. You will need to link VCACHE_LIBS with all the client programs
|
|
* that want to use the VCACHE library.
|
|
*
|
|
* #define INCREASE_CLIENT_EXECUTABLE_SIZE_DO_NOT_EXEC_VCACHE
|
|
*/
|
|
|
|
|
|
|
|
/* Location of wais sources, note trailing slash. MITRAism.
|
|
*
|
|
* Note that it is not clear to me (swa) why this is useful to have for the
|
|
* clients. Isn't this a server-specific thing? It's on my TODO list.
|
|
*/
|
|
#define WAIS_SOURCE_DIR "/usr/local/wais/wais-sources/"
|
|
|
|
/*
|
|
* Filters are not supported in the default distribution because they
|
|
* currently only work on the VAX. We are working on resolving these
|
|
* problems.
|
|
*/
|
|
#define P_NO_FILTERS
|
|
|
|
/* If you want to use the UNIX compatability library (documented in the
|
|
* Prospero Library Reference Manual), remove the definition of
|
|
* P_NO_PCOMPAT and follow the additional directions in the INSTALLATION file.
|
|
* This option is currently slightly broken, because Prospero has undergone
|
|
* a number of changes. it can easily be reported.
|
|
*
|
|
*/
|
|
#define P_NO_PCOMPAT
|
|
|
|
/*
|
|
* This option exists for sites that have special applications. If this option
|
|
* is defined, the pcompat library's open() routines will prompt for an FTP
|
|
* username and password to retrieve files, if necessary. This may increase
|
|
* the amount of information available through applications linked with the
|
|
* compatability library, but at the cost of changing the user interface
|
|
* to applications such as 'cat' which never normally prompt for a password.
|
|
* Only #define this if you have a special application.
|
|
*
|
|
* #define PCOMPAT_SUPPORT_FTP
|
|
*/
|
|
|