This document attempts to answer frequently asked questions on both the AccessNFS DiskAccess NFS client and AccessNFS Gateway. Many of the questions and answers pertain to both products but there are some questions and answers that are product specific. The product specific questions and answers are located in their own sections. The sections are named "Common Questions and Answers", "DiskAccess Questions and Answers", and "AccessNFS Gateway Questions and Answers".
Common Questions and Answers
1. How do I know if I have the latest version?
Use the About utility to determine the version of the product you are running. Connect to the Shaffer Solutions Web site and check for the latest version that is available. To connect to the Shaffer Solutions Web sit, visit http://www.ssc-corp.com/nfs and select "Product Updates".
2. I receive a DiskAccess Service Failure (or NFS Login Failed) after logging
into Windows.
The DiskAccess Service Failure message box which reads, "An error occurred while attempting to communicate with the DiskAccess service. NFS will use the anonymous UID (-2), GID (-1), and factory default mount options." can indicate an error has occurred or it can indicate an authorization problem has caused NFS to simply use the default settings.
In most cases, this message indicates NFS could not authenticate the user who is logging onto the system and having their persistent NFS connections established. The user should first determine if the DiskAccess service is running on the local system. The services control panel utility or the services snap-in located in a Microsoft Management console will list all of the registered services and the corresponding status. The status should read "Started". If status is not "Started", then further investigation of why the service failed to start needs to take place. If assistance is needed, please contact Shaffer Solutions support services.
If authentication is being established with a PCNFSD service, the DiskAccess service may not have been able to contact the server running the pcnfsd service. Attempts should be made to contact the server with the ping.exe utility followed by the RPC Information utility. If the DiskAccess service is running and the ping and rpc information test are successful, then further investigation of why the authentication failed needs to take place. If assistance is needed, please contact Shaffer Solutions support services.
If authentication is being established with NIS, the DiskAccess service may not have been able to contact a NIS server serving the user's NIS domain. The DiskAccess service relies on a NIS server to respond to a NIS broadcast which is used to discover any NIS servers of the given NIS domain. If no NIS servers respond to the broadcast or if a NIS server which has responded to the broadcast goes off line, then the above error message will be displayed. The network administrator will need to confirm the NIS server is working correctly. If the NIS server is working correctly and the DiskAccess services is running, then further investigation needs to take place. If assistance is needed, please contact Shaffer Solutions support services.
3. When I install the product, the Installation Options dialog shows "This
choice will use 0 KB of space on your disk" for the compact option. That
doesn't make any sense.
The installation program first calculates the amount of space needed for each file to be delivered with the installation option you selected. Then it looks to see if any of the files already exist in the installation directory on your system. If so, it subtracts the size of the existing file from the size of the new file. If files in the version you are installing are smaller than the older version currently on your system, the message could even show the disk space needed as a negative number.
4. The file/directory name which I created appears to be truncated when I view
the contents of the containing directory. What happened?
Although the product is capable of supporting long file/directory names, the server may not support long file/directory names. In many cases, the server may truncate the name when creating or renaming files and directories. This is particularly true when using Intergraph Corporation servers running Intergraph's CLIX operating system.
5. How does the product obtain the list of NFS servers displayed when the user
clicks on NFS Network from Network Neighborhood?
The product's Network Provider DLL periodically issues a network broadcast to request a reply from each NFS server in the configured NFS LANs as defined in the "Configured NFS LANs" tab of the product's configuration property sheet. The name of each server that responds to this broadcast is placed in a list that pertains to the specific NFS LAN. This list is displayed when an application browses the specified LAN in the NFS Network.
The product will identify the local network as "DefaultLAN" by default. The system administrator can establish a NFS LANs by using the product's configuration property sheet. The tab entitled "Configured NFS LANs" should be used to establish other NFS LANs. Please refer to the online help for additional information on configuring NFS LANs.
By default, the product's Network Provider issues a broadcast at system startup
and again when the refresh interval as indicated on the "Configured NFS
LANs" expires. A new list of servers are generated with each broadcast.
In this manner, servers that are no longer responding are eliminated from their
respective NFS LAN list, and new servers are added to the list.
The product may be configured to only issue the broadcast once, at system startup.
This method reduces the amount of network traffic. However, the NFS server list
for the corresponding NFS LAN will never be updated and, therefore, may become
inaccurate if servers are removed from or added to the network.
The product may be configured to only issue the broadcast when an application attempts to browse a specific NFS LAN in the NFS Network. This provides a relatively up-to-date list of NFS servers and generates the least amount of network traffic. The first time a user browses a NFS LAN configured in this fashion, the product will broadcast to the NFS LAN. If the user attempts to browse the same NFS LAN before the "refresh interval" expires, the previous list of NFS servers will be provided. However, if the "refresh interval" has expired, the product will broadcast to the NFS LAN, regenerate the list and restart the "refresh interval" timer. However, this technique can cause the browsing application to wait for the list to be generated.
If the user can refresh the internally stored lists of NFS servers that correspond to the configured NFS LANs by either depressing the "Refresh Now" or "Refresh All" buttons on the "Configured NFS LANs" tab of the product's configuration property sheet. The "Refresh Now" will only update the list of NFS servers for the NFS LAN which is currently highlighted in the "NFS LANs" listbox. The "Refresh All" will update all lists for all of the NFS LANs which are listed in the "NFS LANs" listbox.
There is a configurable time-out value that specifies the length of time the product's Network Provider will wait to receive replies from NFS servers responding to the broadcast. By default, this time-out value is set to 15 seconds. This long time-out period allows servers that are slow to respond, such as servers on distant LANs, to be included in the list. If the product is configured to only broadcast upon demand, it is recommended that this time-out value be set to one second. Although this will prevent slow responders from being added to the list, this will improve the browse response time.
6. Where can I get PCNFSD for my server system?
PCNFSD has been ported to many different platforms. The following is a partial list of sites for different server platforms:
SunOS 4.x, Solaris bcm.tmc.edu
Solbourne src.doc.ic.ac.uk
NeXTStep ftp.york.ac.uk:/pub/pv/pc-nfs/RPC.pcnfs/*
Ultrix 4.2 bcm.tmc.edu
IRIX/SYSV ftp.sgi.com:/support/pcnfsd.sysV [unsupported]
IBM AIX 3.2 Call IBM and ask for PTF# U412556
IBM AIX 3.2.1 Call IBM and ask for PTF# U419359
IBM AIX 3.2.3 Call IBM and ask for PTF# U414701
IBM MVS Call IBM and ask for PTF# UY84244 [pcnfsd v1 only]
MIPS ftp.york.ac.uk:/pub/pc/pc-nfs/RPC.pcnfsd/*
Many more ftp.york.ac.uk:/pub/pc-nfs/RPC.pcnfs/pcnfsd.tar.
7. Do you have any examples of setting up PCNFSD on a server?
Examples are included below for the following NFS servers:
Windows NT Servers running Shaffer Solutions DiskShare
SunOS 4.x Operating System
Solaris 2.x (SunOS 5.x) Operating System
Windows NT Servers running Shaffer Solutions DiskShare
The Shaffer Solutions DiskShare product lets Windows NT systems act as NFS servers, making directories and printers available to NFS clients. All printers that are configured on a system running DiskShare are available to clients using PCNFSD printing; however, the printers cannot be browsed because DiskShare does not implement PCNFSD version 2 completely. PC-NFS clients will need to specify the server name and printer name explicitly to connect to a printer on a DiskShare server.
If you want the PC-NFS daemon (PCNFSD) to start automatically on a DiskShare server when the system is booted, perform the following steps:
1. In Control Panel, open the Services icon.
2. From the list of available services in the Services control panel, select pcnfsd;
then select the Startup button.
3. In the Service dialog that is displayed, select a Startup Type of Automatic.
Selecting Manual will require you to start PCNFSD manually each time you
reboot the system.
4. Close the Service dialog; then close the Services control panel.
SunOS 4.x Operating System
It is recommended that the user refer to the SunOS documentation for LPR and the delivered manual page for rpc.pcnfsd(8) for more complete instructions on how to configure PCNFSD on the SUN Operating System. The following is a brief summary of the steps required to load the rpc.pcnfsd file on the Sun machine:
1. Copy the file rpc.pcnfsd to /usr/etc/rpc.pcnfsd
2. Make sure the permissions and ownership of the file are as follows:
3. -rwxr-xr-x 1 root 65536 Mar 17 1994 /usr/etc/rpc.pcnfsd
4. Add the following to the /etc/rc.local file after the entry for ifconfig -a:
#
if [ -f /usr/etc/rpc.pcnfsd ]; then
/usr/etc/rpc.pcnfsd -s /usr/spool/pcnfs;
echo 'rpc.pcnfsd' > /dev/console
#
5. Create the following file:
/usr/spool/pcnfs
Authentication: When pcnfsd receives a PCNFSD_AUTH or PCNFSD2_AUTH request, it will "log in" the user by validating the username and password and returning the corresponding user ID, group IDs, home directory, and default file permissions (umask). At this time, pcnfsd will also append a record to the /etc/wtmp database. To record PC "logins" in this way, you should add a line with the following format to the pcnfsd.conf file:
wtmp yes (The pcnfsd.conf file is delivered with the value of wtmp set to "no.")
Printing: The pcnfsd daemon supports a printing model that uses NFS to transfer the actual print data from the client to the server. The client system issues a PCNFSD_PR_INIT or PCNFSD2_PR_INIT request, and the server returns the path to a spool directory which is exported by NFS and which the client may use. The pcnfsd daemon creates a subdirectory for each of its clients: the parent directory is normally /usr/spool/pcnfsd and the subdirectory is the hostname of the client system. To use a different parent directory, add a line with the following format to the pcnfsd.conf file:
spooldir <path> (i.e. spooldir /home/spool)
Once a client has mounted the spool directory using NFS and has transferred print data to a file in this directory, it issues a PCNFSD_PR_START or PCNFSD2_PR_START request. The pcnfsd daemon handles this, and most other print related requests, by constructing a command based on the printing services of the server operating system and executing the command using the identity of the PC user. Since this involves set-user-id privileges, pcnfsd must be run as root.
Every print request from the client includes the name of the printer that is to be used. To pcnfsd, a printer appears as either a destination serviced by the system print spooler, or as a virtual printer. Refer to the SunOS lpc(8) manual page for more information on setting up the system print spooler to handle a new printer. Virtual printers, known only to pcnfsd clients, are defined in the /etc/pcnfsd.conf file by a line with the following format:
<printer name> <alias-for> <command>
The printer name parameter is the name of the printer you want to define. The alias-for parameter specifies the name of a "real" printer which corresponds to this printer. Up to sixteen virtual printers can be defined. For example, a request to display the queue for printer name will be translated into the corresponding request for the printer alias-for. If you have defined a printer in such a way that there is no "real" printer to which it corresponds, use a single dash (-) for this field. (See the example for the printer named "test" below.) The "command" parameter will be executed whenever a file is printed on the defined printer name. This "command" is executed by the Bourne shell using the -c option. For complex operations you should construct an executable shell program and invoke it as a "command". Within the "command" parameter, the following tokens will be replaced dynamically by pcnfsd:
$FILE Replaced by the full path name of the print data file. When the
command has been executed, the file will be unlinked.
$USER Replaced by the user name of the user logged into the client system.
$HOST Replaced by the host name of the client system.
$PRINTER Replaced by the real printer name.
Consider the following example pcnfsd.conf file:
printer aliasfuji fuji lpr -P fuji -J $USER -C $HOST $FILE
printer test - /usr/bin/cp $FILE /usr/tmp/$HOST-$USER
If a PC client system prints a job on the printer "aliasfuji", pcnfsd will execute the specified print job on the printer "fuji". If the client requests a list of the print queues for the printer "aliasfuji", the pcnfsd daemon will translate the request for a listing of the printer "fuji" .
The printer "test" is used only for testing. Any file sent to this printer will be copied into /usr/tmp. Any request to list the queue, check the status, etc... of printer "test" will be rejected because the alias-for has been specified as "-".
Solaris 2.x (SunOS 5.x) Operating System
It is recommended that the user refer to the Solaris documentation for LP and LPR, and the manual page delivered with the rpc.pcnfsd(1M) daemon for more complete instructions on configuring PCNFSD on Solaris (SunOS 5.x) systems. The following is a brief summary of the steps required:
To install pcnfsd:
1. Copy the file rpc.pcnfsd to /usr/lib/nfs
2. Execute a 'chmod 555', 'chown bin', and 'chgrp bin' on the above file.
3. Start the daemon by typing /usr/lib/nfs/rpc.pcnfsd
4. Edit the /etc/init.d/nfs.server file for automatic start/stop of the daemon
***.
*** One line needs to be added, a second line needs to be edited.
First, find the line that shows
/usr/lib/nfs/nfsd -a 16and add this line below it
/usr/lib/nfs/rpc.pcnfsd
Then, find the line that shows
'(nfsd|mountd|rpc.pcnfsd|rpc\.bootparamd|in\.rarpd|rpld)'and edit it so that it looks like the one below
'(nfsd|mountd|rpc.pcnfsd|rpc.pcnfsd|rpc\.bootparamd|in\.rarpd|rpld)'
Authentication:
When pcnfsd receives a PCNFSD_AUTH or PCNFSD2_AUTH request, it will "log in" the user by validating the username and password and returning the corresponding user ID, group IDs, home directory, and default file permissions (umask). At this time, pcnfsd will also append a record to the /etc/wtmp database. To record PC "logins" in this way, you should add a line with the following format to the pcnfsd.conf file:
wtmp yes (The pcnfsd.conf file is delivered with the value of wtmp set to "no") .
Printing: The pcnfsd daemon supports a printing model that uses NFS to transfer the actual print data from the client to the server. The client system issues a PCNFSD_PR_INIT or PCNFSD2_PR_INIT request, and the server returns the path to a spool directory which is exported by NFS and which the client may use. The pcnfsd daemon creates a subdirectory for each of its clients: the parent directory is normally /usr/spool/pcnfsd (for the LPR print system) or /usr/spool/lp/pcnfs (for the LP print system) and the subdirectory is the hostname of the client system. To use a different parent directory, add a line with the following format to the pcnfsd.conf file:
<spooldir> <path> (i.e. spooldir /usr/printspool)
Once a client has mounted the spool directory using NFS and has transferred
print data to a file in this directory, it issues a PCNFSD_PR_START or PCNFSD2_PR_START
request. The pcnfsd daemon handles this, and most other print related requests,
by constructing a command based on the printing services of the server operating
system and executing the command using the identity of the PC user. Since this
involves set-user-id privileges, pcnfsd must be run as root.
Every print request from the client includes the name of the printer that is to be used. To pcnfsd, a printer appears as either a destination serviced by the system print spooler, or as a virtual printer. Refer to the lpadmin(1M) or lpc(1B) manual pages for more information on setting up the system print spooler to handle a new printer. Virtual printers, known only to pcnfsd clients, are defined in the /etc/pcnfsd.conf file by a line with the following format:
<printer name> <alias-for> <command>
The printer name parameter is the name of the printer you want to define. The alias-for parameter specifies the name of a "real" printer which corresponds to this printer. Up to sixteen virtual printers can be defined. For example, a request to display the queue for printer name will be translated into the corresponding request for the printer alias-for. If you have defined a printer in such a way that there is no "real" printer to which it corresponds, use a single dash (-) for this field. (See the example for the printer named "test" below). The "command" parameter will be executed whenever a file is printed on printer "name". The "command" is executed by the Bourne shell using the -c option. For complex operations you should construct an executable shell program and invoke the shell program via "command". Within "command" the following tokens will be replaced dynamically by pcnfsd:
$FILE Replaced by the full path name of the print data file. When the
command has been executed, the file will be unlinked.
$USER Replaced by the user name of the user logged into the client system.
$HOST Replaced by the host name of the client system.
$PRINTER Replaced by the real printer name.
Consider the following example pcnfsd.conf file:
printer aliasfuji fuji lpr -P fuji -J $USER -C $HOST $FILE
printer test - /usr/bin/cp $FILE /usr/tmp/$HOST-$USER
If a PC client system prints a job on the printer "aliasfuji", pcnfsd
will execute the specified print job on the printer "fuji". If the
client requests a list of the print queues for the printer "aliasfuji",
the pcnfsd daemon will translate the request for a listing of the printer "fuji".
The printer "test" is used only for testing. Any file sent to this
printer will be copied into /usr/tmp. Any request to list the queue, check the
status, and etc... of printer "test" will be rejected because the
alias-for has been specified as "-".
8. When I copy a CD-ROM to the server, I can't create files in the new directory. This sometimes happens with other directory trees as well. My user credentials seem okay, so what is wrong?
If the directory tree you copied is read-only or has read-only subdirectories, the attributes on the directory you create are set to read-only access, so you can't create files. If you change access on the new directory to read-write you will be able to copy the files into this directory. You must repeat this process for each read-only directory on the source media.
9. Why can I not mount the root partition "/", from my UNIX NFS server
to my Windows system?
As of version 3.2.X.X, the root partition "/", of a NFS server can be mounted. However, special syntax must be used. The exclamation point "!" is used to indicate the desire to mount the NFS server's root partition. For example, to mount the root partition on the NFS server named nfs_server, the syntax used in Explore's Path field in the Mapped Network Drive dialog box would be:
\\nfs_server\! (if you prefer to use Windows notation) or
nfs_server:/! (if you prefer to use UNIX notation).Only use the exclamation point to indicate the need to mount the root partition. Do not use it in place of the first forward slash when attempting to mount directories that are child directories of the root partition.
10. My NFS server is exporting both /usr and /usr/projects, which are two different
disks. Why can I not see the files and directories in the /usr/projects directory
if I mount only the /usr directory?
An NFS server with an export list like the following presents Windows with name ambiguities:
/usr
/usr/projects
Windows and AccessNFS product cannot determine whether "/usr/projects" refers to the _directory_ "projects" under the "/usr" export or it refers directly to the "/usr/projects" _export_.
Under these conditions, the product will only allow access to either the directory "projects" in the "/usr" directory or the "/usr/projects" mount point, depending on how the resources are mounted. If you wish to access "/usr/projects", the mount point, or any directories under the mount point "/usr/projects", then you must perform those mounts prior to mounting the "/usr" export. General rule: If the server presents ambiguous NFS resources, then mount the longest NFS resource name first.
Once you have mounted the longest ambiguous path, you can then mount the shorter ambiguous paths. However, when accessing the ambiguous name (e.g. the "projects" directory under "/usr" mount point), the product will resolve the longer name of "/usr/projects", which is a mount point.
The UNC name "\\server\usr\projects", is handled differently depending on the requesting application. Some applications request to mount just the "\usr", while other applications request to mount the entire name, "\usr\projects". To keep all applications consistent, it is recommended that you create a persistent connection to "\usr\projects" using a drive letter.
A side effect of the above scenario is one which the product will allow the user to mount the /usr directory of a NFS server, given the fact the NFS server is only exporting the directory /usr/project. The user will not be able to perform any options in the /usr directory other than perhaps reading or writing to the /usr/project directory, assuming the user has the necessary file access privileges in /usr/project.
11. I cannot update files using the Microsoft Office 97 utilities. When I try
to update the files and save the changes, I receive a network error. However,
other Microsoft utilities, such as notepad and wordpad work correctly.
Due to the way some applications will automatically change the case of filenames regardless of how the user types the name, it can cause problems when attempting to write files to NFS servers whose underlying file system is both case preserving and case sensitive. To avoid the problem, DiskAccess provides filename creation and lookup options. The options can be found on the "Filenames" property tab of the DiskAccess property sheet. Set the options "Convert to Lower Case", "Match Filenames Exactly", and "Match Filenames Ignoring Case".
12. If multiple people attempt to access the same PowerPoint file with PowerPoint
SR-1 from a NFS drive with locking turned on, all of the instances of the file
are opened as read-only; even the first instance of the opened file.
PowerPoint relies on the underlying file system to notify it whenever a file which has been locked by another user is available for write access. Since NFS is a stateless protocol, it does not know of the concept of file system change notification. Consequently, if multiple people access the same PowerPoint file on NFS drives which have locking turned on, they will all open the file as read-only.
13. I cannot mount the NFS resouce /IPC$ on my NFS server.
Windows will often send file system requests to the AccessNFS driver with requests to mount /IPC$, /IPC$/srvsrc, or /IPC$/spooler. In the context of Windows Networking, they are used as a means for interprocess communication. NFS does not support the concept of interprocess communication. Consequently the AccessNFS driver will not send a mount request for any resources which start with "/IPC$" or "/IPC$/abc123" where abc123 could be any word.
14. Do the DiskAccess and Gateway products have command line utilities?
Yes, both products have utilities that can be run in a DOS window. All of the command line utilities have an equivalent graphical counterparts. See the online help for details on the command line options to run the command line utility with the "/?" option and an usage statement should be given.
ProductDir is defined as the directory name where the product has been installed. For example, c:\Program Files\DiskAccess
DiskAccess:
| Name | Location | Description |
| daconfig.exe | ProductDir\bin | Configure default NFS information |
| nfs.exe | ProductDir\bin | Mount/unmount NFS resources |
| ntp.exe | ProductDir\NetSvc | Network Time Protocol client |
| rpcinfo.exe | ProductDir\Utilities | ONC RPC server locator |
| showmount.exe | ProductDir\Utilities | Enumerates NFS Server's exported directories |
| dig.exe | ProductDir\wdig | Domain Name System (DNS) look-up client |
| finger.exe | ProductDir\wfinger | Displays user information via finger protoco |
Gateway:
| Name | Location | Description |
| gwshare.exe | ProductDir\Admin | Create and delete shared NFS drives |
| mapusers.exe | ProductDir\Admin | Manage Windows and NFS user mappings |
| rpcinfo.exe | ProductDir\Utilities | ONC RPC server locator |
| showmount.exe | ProductDir\Utilities | Enumerates NFS Server's exported directories |
| dig.exe | ProductDir\wdig | Domain Name System (DNS) look-up client |
15. Clear up my confusion about the "Existing Filename Case" options
in the Filenames tab sheet. How can you match filenames exactly and ignore case?
"Match Filenames Exactly" is the only option select
This option has dependencies on the option chosen in the "New Filename Case". If the "Preserve Case" option has been selected, then no case conversion is performed on the file or directory name prior to searching for the file or directory on the NFS server. In other words, DiskAccess/Gateway will use the name as it was given by the application. Keep in mind that some application will automatically convert the file or directory name to upper case even though the user may have entered the name in lowercase or a mix of case. In this event, users often mistaken believe the product is not working correctly.
If either the "Convert to Lower Case" or "Convert to Upper Case" is the only option select then DiskAccess/Gateway will either lower case or upper case the entire file/directory name depending on the option that has been selected prior to conducting the search for the file/directory name on the NFS server.
"Match Filenames Ignoring Case" is the only option select
There are no dependencies to the "New Filename Case" selected options when "Match Filenames Ignoring Case" is the only option selected. DiskAccess/Gateway will lower case the file/directory that is being sought as well as the list of file/directory names of the parent NFS directory and then the search will take place. This is truely a case insensitive search.
Both "Match Filenames Exactly" and "Match Filenames Ignoring Case" are selected
DiskAccess/Gateway will follow the same routine as when only the "Match Filenames Exactly". However, if a match is not found then DiskAccess/Gateway will then use the routine as described in the "Match Filenames Ignoring Case" section.
The main reason for the options and some of the confusion is because in large part the Microsoft Windows NT and Windows 2000 operating systems are case preserving but not case sensitive. If you want DiskAccess/Gateway to act like UNIX or Linux and you have smart applications, then you will want to turn on the "Preserve Case" and "Match Filenames Exactly" options.
16. Why can I still browse an NFS server that I know has just gone off-line?
Once an NFS server has been browsed, the list of exported directories are stored in memory. By default, the stored information will remain valid and accessible for two minutes. After the default time out value of two minutes any new requests for the server's exported resources will result in a network to the server for the information. The new or updated information will then be stored in memory.
Since the information is stored in memory, the stored information can become inconsistent with the actual exports list of the server if by chance the server's list changes within the timeframe of the memory time out. The time out value can be configured by changing a registry data value. The registry value is ExportsCacheInterval and is located in the registry key,
HKEY_LOCAL_MACHINE\SOFTWARE\SSC\DiskAccess\CurrentVersion\Default
The minimum allowed data value is 1 and the maximum data value is 1440. The value represents minutes that a server's export list will remain in memory.
1. I have enabled locking on all of my DiskAccess NFS clients, but when I try to access a file with two different clients, I get no warnings or errors. How can this be?
In addition to enabling locking on the clients, locking must be available and enabled on the server being accessed. The lock manager on the server must also support Network Lock Manager version 3 or greater because this is the only version that supports PC type lock requests.
2. I have enabled locking on my DiskAccess NFS client and now it runs very slowly.
Why is this?
Enabling locking will slow the client down in any case because of the extra network traffic required to support locking. There are cases where the slow down could be extreme, however. There are two reasons this could happen. Either locking is not enabled on the server being accessed, or the server does not support Network Lock Manager version 3 or greater. In both of these cases, the client is attempting to contact the lock manager (version 3) on every access, timing out, and retrying. This behavior is required if the client is to recover automatically from a server crash.
3. Application XYZ (or everything) has slowed down. Why? It takes me 3 seconds
to do anything. Why? I can never get mounted to certain NFS servers. Why?
In the registry there is a value called "FirstContact" under the key:
HKEY_LOCAL_MACHINE\SOFTWARE\SSC\DiskAccess\CurrentVersion\Default
The number contained in "FirstContact" is the time to wait for an initial response when attempting to contact a new machine. The value is in tenths of a second and has a default value of 16 (1.6 seconds). The default value may be set to 3 on some systems due to older versions of the product; however, if set to 3 it will automatically be translated into 1.6 seconds. If a machine has been removed from the network or does not support NFS, it will take 1.6 seconds for DiskAccess to error out.
If you have placed a large number of (now defunct) UNC names in your path, e.g. PATH=\\oldnode1\path;\\deadnode\path123;\\lamebox2\bindir;c:\winnt, then DiskAccess will time-out on each UNC name (1.6 seconds each) for every time the command prompt searches your path. But it gets worse. Windows Networking tries to resolve the UNC name first. If the name of the server in the UNC path is a fully qualified domain name, then Windows Networking takes over 5 seconds to determine if the server supports Windows Networking. If the server's name is not fully qualified, then Windows Networking can take longer than 5 seconds, depending on how the system name to address resolution is configured (i.e. DNS, WINS, or local look up).
A similar problem can happen to any application with a stale UNC name. If you mistype the nodename in the "Map Network Drive" box or have a stale Windows shortcut ("\\deadnode\netpath"), "FirstContact" determines the amount of time before DiskAccess gives up trying to contact the non existent server. The "FirstContact" time-out added to the Windows Networking time-out will significantly slow the system down.
The best solution is to remove invalid UNC names from your PATH and applications. If removal of the UNC names are not possible, then you can tune the value of "FirstContact" to time-out quicker.
Note that a small value for "FirstContact" may not allow NFS connections to remote nodes which reside on the Internet, accessible across a slow link, or to NFS servers that are slow to respond due to a heavily used local network.
4. I have DiskAccess loaded on my laptop computer and use it to dial in to the
office. I have to wait about 4 minutes during login before DiskAccess pops up
a message box saying that the NFS login has failed. How can I get rid of this
delay?
The delay occurs because DiskAccess is trying to contact the PCNFSD or NIS server for NFS login authentication, but it is not yet available. Eventually, a time-out will occur and DiskAccess will return the login failure message. To avoid the delay, invoke the DiskAccess configuration property sheet from the Control Panel and remove the authentication server name. If there is no server name present, DiskAccess will immediately return a login failure without attempting to contact the server. Consequently this means that all subsequent NFS logins that use the configured base defaults will fail.
5. I don't like the solution to question number 4. I use DiskAccess on my computer
at home, which also functions as a fax modem. If there is a power failure and
the system reboots, login is now halted until someone comes along and dismisses
the login failure box. Is there a way to get rid of it altogether?
Yes. DiskAccess can be configured so that NFS login is not attempted until an NFS mount is attempted. However, this means that browse connections (those done through Network Neighborhood without an associated drive letter) will use the factory defaults rather than the configured base defaults. To configure DiskAccess to bypass NFS login at system login, invoke the DiskAccess configuration property sheet from the Control Panel and clear the Authenticate At System Logon check box.
6. I have a server running NIS+ and my NIS domain names that have the form of
"yp.wright.gov". DiskAccess NIS Authentication doesn't seem to work.
What is the problem?
As far as DiskAccess NIS Authentication goes, it does not do any syntax checking
on the domainname that a user has typed. Since DiskAccess is using the NIS protocol
to communicate with the server, the NIS+ server simply will not respond. The
server needs to be configured so that it runs in NIS Compatibility mode.
7. What are credentials? Which credentials would I use if I am browsing the
Network Neighborhood?
Default user credentials can be set only through the product's Configuration
utility from the Control Panel after manually logging into Windows. Each mapped
drive can have alternate credentials by configuring through the DiskAccess Configuration
property sheet while mapping the network drive, but the alternate credentials
will not be saved other than for that mapped drive.
UNC connections will use default user credentials, including browsing the Network Neighborhood, and thus; if you would like to run a service using a particular user's credentials, you will have to manually log in as the user and set the default user credentials for the user through the DiskAccess Configuration dialog in the Control Panel.
If you haven't configured your Default user Credential or failed to configure during login, your UID and GID will be -2 and -1, respectively. While using OLE applications, all mapped drive letters will revert back to UNC names, and UNC connections WILL NOT carry the alternate credentials which were set for the mapped drive. They will use the default user credentials.
Lastly, persistently mapped drives that have been mounted with the default credentials will continue to use the default credentials as they existed during the initial mapping of the drive, regardless of future changes to the default user credentials.
8. In PC-NFS for NT, I could set the default options to preserve case for creating
files and ignore case for finding files. DiskAccess will not allow me to choose
Match Filenames Ignoring Case for existing files if I select Preserve Case for
new files. I really need this functionality. Is there any way to force the options
I want?
Yes. You can manually force DiskAccess to read the entire directory and match filenames regardless of case, even when the Preserve Case option has been selected. WARNING: This mode is NOT RECOMMENDED. Some applications do not preserve the filename case that you type into the SaveAs dialog, or they append a different case extension, etc. If files differing only in case exist on the server directory, then these applications may unexpectedly overwrite the wrong file. You should only enable Match Filenames Ignoring Case when Preserve Case is set if you fully understand the potential consequences and have control of filenames on the server.
As of version 3.2.X.X of the product, whenever Preserve Case has been selected, you can now set Match Filenames Ignoring Case.
9. Why can't system services which impersonate the logged on user access files
and directories on the NFS mounted drive that are owned by the logged on user?
The logged on user has not established their default NFS credentials. Consequently, any service which attempts to access NFS mounted drives as the logged on user will do so with the factory default credentials. The factory default credentials are set as the NFS user "nobody". On most NFS servers, the user "nobody" only has access rights to those files and directories which are open to everybody. For example, on a UNIX NFS server, the user "nobody" could only access files whose world permission bits are set to 7 or 6.
To establish the logged on user's default NFS credentials, the logged on user must use the DiskAccess configuration property sheet. Online help is available with the DiskAccess configuration property sheet. Once the logged on user establishes their default NFS credentials, any system service which impersonates the logged on user will use those NFS credentials which were established with the DiskAccess configuration property sheet.
10. Why can system services which are logged on as LocalSystem only access files
and directories on the NFS drive which are open to everybody?
The answer to this question is similar to that of the previous question. The default NFS credentials have not been established for the user account, LocalSystem. However, a user cannot simply log into the Windows NT as LocalSystem and use the DiskAccess configuration property sheet to establish the default NFS credentials. DiskAccess does provided another utility that can only be run by user who is a member of the administrators group. This utility is called "Administrator Utility". When the Administrator Utility is started, it will execute the DiskAccess configuration property sheet in it's own unique environment. Consequently, the default NFS credentials established will only be for the special user account, LocalSystem.
Once the default NFS credentials have been established for LocalSystem, all system services which are logged on as LocalSystem will access NFS mounted drives with the same NFS credentials.
11. DiskAccess does not honor the "block and wait" characteristics
of functions such as LockFileEx() with the LOCKFILE_EXCLUSIVE_LOCK flag.
The current version of DiskAccess does not honor the "block and wait" locks. If the file has not been locked by another process, the requesting locking process will be granted the lock. However, if the file is currently locked by another process and a requesting process attempts to lock the file, LockFileEx() will fail with the error ERROR_LOCK_FAILED, instead of blocking and waiting for the other process to release the lock.
12. PCNFSD user authentication fails whenever I use the PCNFSD server which
is located on another campus. How can the timeout value be increased?
In the registry there are two values which are used. The name of the values
are "FirstContact" and "Retransmissions". The values are
located in the key:
HKEY_LOCAL_MACHINE\SOFTWARE\SSC\DiskAccess\CurrentVersion\Default
The number contained in "FirstContact" is the time to wait for an
initial response when attempting to contact the system running the PCNFSD service
over TCP. The value is in thousandths of a second and has a default value of
1600 (1.6 seconds). This value can be increased above 1600 to accommodate slow
PCNFSD servers; however, it is recommended not to set the value lower than 1600.
The "Retransmissions" value is used whenever the system is unable to acquire a local TCP port and a UDP port has to be used instead. Due to the nature of UDP, applications may need to retransmit packets. The "Retransmissions" value sets the maximum number of retransmissions allowed before the request is timed out.
13. DiskAccess on Windows NT and Windows 2000 and AccessNFS Gateway will only
display symbolic links that can be resolved. How can unresolved symbolic linked
files be displayed?
If the "Resolve Symbolic Links" option has been turned on in the "Symbolic Links" section of the control panel utility, then all symbolic links in a directory will be displayed provided the system is able to resolve the symbolic link to a target file or directory. For the case in which the target file or directory cannot be resolved, the symbolic link will not be displayed. If unresolved symbolic linked files must be displayed you can change this behaviour using the ControlPanel-DiskAccess Configuration-Symbolic Links tab
14. DiskAccess on Windows NT and Windows 2000 and AccessNFS Gateway do not allow
the user to delete or rename a symbolic link. How can the actual symbolic link
be deleted or renamed?
If the "Resolve Symbolic Links" option has been turned on in the "Symbolic Links" section of the control panel utility, then all symbolic links in a directory will be displayed provided the system is able to resolve the symbolic link to a target file or directory. A user is not allowed to rename or delete a symbolic link by default. However, if the user wishes to delete or rename the actual symbolic link you can change this behaviour using the ControlPanel-DiskAccess Configuration-Symbolic Links tab.
Users must be aware of how some Windows applications will create temporary files for editing. After the user selects the option to save the file, the application will replace the actual file with the newly edited temporary file. If the file was located on a NFS drive and was also a symbolic link, the 'save' operation will often fail because the Windows application will not be allowed to replace the symbolic linked file. If the DeleteSymlinks value is set, the operation will succeed; however, the symbolic link will now be replace with an actual file. Consequently, the user will have two copies of the file on the NFS server; the newly edited file in place of the symbolic link and the previous version of the file located in the now defunct symbolic link's target directory. Applications which do not create temporary files should not have any difficulties access files on NFS drives which are actually symbolic linked files.
12. What happens to a user's persistent NFS connections when the system is rebooted
and the user's NFS password changes?
Once the user logs onto the system, all persistent network connections will be re-established. However, if the user's NFS password has changed, then the user will be prompted to enter their new password. In Windows NT 3.51, the user will be presented with the DiskAccess control panel utility. In Windows NT 4.0, the user will be presented with a dialog box titled "Enter Network Password", with the caption "Incorrect password or unknown user for: <name of network resource>".
13. When browsing in the Network Neighborhood, if I look at the properties of
a file or directory located on a NFS server, there is no filename or icon at
the top of the General tab sheet.
This problem occurs when looking at the properties for files and directories located in an export which has multiple path components. An example would be the file "readme.txt" located under the NFS export "/usr/tmp". The filename and icon would not be displayed at the top of the General tab for this file. If this file were located under the NFS export "/usr", the filename and icon would be correctly displayed. This problem only occurs in the Network Neighborhood. It does not occur when a drive letter has been mapped to the server. This problem is caused by a limitation in the operating system.
14. After dragging a portion of text from Wordpad and dropping it on top of
a NFS subdirectory and thus creating a scrap/shortcut document, I am unable
to access the document located on the NFS subdirectory. Why?
The above problem is a result of not enabling the option, "Provide Unique 8.3 Format Names" at mount time and the resulting name of the file is not in the 8.3 format. The 8.3 format being a name with eight or less characters followed by a dot "." followed at least one character but no more than three characters. An example of an 8.3 formated name would be, football.txt.
Often the resulting error message seen by the user will be, " The name <drive>:\<subdirectory_name>\ is not valid.".
15. DiskAccess running on Windows NT creates extra files which end with ":afp_afpinfo"
whenever I copy files from my local drive to the NFS mounted drive.
This problem only occurs when the DiskAccess system currently has or once had
Microsoft's "Services for Macintosh" loaded. Furthermore the problem
occurs when copying files from local directories which reside or once resided
in a Macintosh-Accessible volume. For example, if the file dan.txt resides in
a Macintosh-Accessible volume and a user copies the file to a NFS mounted drive,
the file dan.txt and dan.txt:afp_afpinfo will be created on the NFS server.
The operating system informs DiskAccess to create both dan.txt and dan.txt:afp_afpinfo.
If the above example is attempted over a Microsoft network drive via LanMan/SMB the dan.txt:afp_afpinfo file is not created. However, a network packet capturing program does show the creation of the dan.txt:afp_afpinfo file taking place. It is unknown why the LanMan/SMB server does not create the file.
16. Why does the 3Com Fast-IP configuration utility start everytime I attempt
to run DiskAccess' daconfig.exe utility?
If you computer has a 3Com Fast Etherlink XL NIC card in it there is a good chance that 3Com installed a file named %SystemRoot%\System32\daconfig.exe. If your user environment PATH variable has the system32 directory listed prior to DiskAccess' bin directory, then you will execute the 3Com daconfig.exe utility instead of the DiskAccess utility of the same name.
1. Explain the differences between the DiskAccess control panel utility and the "Universal Options" utility in the Gateway menu.
The best way to understand the differences between the two utilities is to remember that AccessNFS Gateway can also be used as a standard NFS client. The "DiskAccess" control panel utility will have the title "AccessNFS Gateway" followed by the user's login name followed by the word "Properties". This utility establishes the logged in user's default authentication credentials and default mount options. The information is used when the logged-in user uses Gateway as a stand-alone NFS client. In other words the user mounts a NFS drive to the system using any number of utilities such as Explorer.exe, net.exe, nfs.exe, or access a NFS server by an UNC path. This same utility often appears when the user logs into the Gateway computer when the product is attempting to acquire the user's NFS authentication information and allow them the opportunity to change some of their default settings.
The "Universal Options" utility has the same appearance as the control panel utility; however, the title will be "AccessNFS Gateway NT AUTHORITY\SYSTEM Properties". This utility is used to establish the default options used by the Gateway Disk Administrator or gwshare.exe command line utility. When the Gateway Disk Administrator is used to mount and share a NFS drive the information supplied to the "Universal Options" utility will appear in the property sheet that will appear after the "Connect" button is selected. Any information placed in the "Authentication" tab sheet is ignored and not used when Windows users access the shared NFS drive.
If the share is being accessed by a service that is running in the context of the local System account, then the authentication and mount credentials that were established with the "Universal Options" utility will be used. This is often the situation the users of Rational's ClearCase product encounter.
Windows users that are accessing the shared NFS drive will use whatever UNIX/Linux
UID/GID pair that was established by using the Gateway "User Manager"
utility or mapusers.exe command line utility. There is one exception to this
rule. The exception is for any Windows user who may have logged onto the Gateway
computer and has established NFS authentication. The established authentication
credentials will override any mapped UID/GID for the user.