Added support for persistent file handles for files on NTFS volumes. This means that you can restart Allegro NFS or the machine that it is running on and not suffer stale NFS file handle errors on the client. Non-NTFS files (or inaccessible files on NTFS volumes) still use non-persistent file handles are still subject to stable NFS file handle errors after restarts.
Allow control over how long file attribute are cached through the configuration GUI. See the “File attribute caching time” setting in the Global tab of the Allegro NFS configuration program.
Configuration for “Host lists” allows host names in the “New address” field. When loading the configuration and resolving the host names, users will be warned if the conversion of saved host names fails, and the host name will be ignored.
When constructing file mode bits for reporting to an NFS client, Allegro NFS uses the filename extension to determine if the file should have the executable (x) bits set. Previously only files with extension .exe, .com, or .bat were marked executable. This list is now adjustable by modifying nfs.cfg and adding an entry for *executable-types*. This operation should only be performed by advanced users.
an NFS client is repeatedly calling stat() on a file to see if its attributes had changed, and
the attributes of the file were altered outside of Allegro NFS’s knowledge.File attributes are cached for a maximum of 5 seconds. Prior to this change, the expiration for cached file attributes would be extended each time they were accessed. This could be a problem in the following scenario: (In this case the new file attributes would never be returned, at least not until the client stopped probing for a sufficiently long time.)
Use higher precision timestamps in log messages, enabled via an option on the debug tab.
Change default statfs blocksize from 8192 to 512. This improves compatibility with some broken NFS clients which do not work properly if the blocksize is not 512.
Minor change to showmount output.
General performance and stability improvements.
Fix: input validation bugs in the Allegro NFS configuration program.
Fix: mounting a subdirectory of an export. For example, if you have an export named “/export” which has a subdirectory named “files”, then mounting servername:/export/files from an NFS client now works properly.
Fix: file and directory changes on the server weren’t seen by some clients. This would make files appear to be static even though they had been modified on the server.
Fix: some previously invisible files are now seen by clients, such as in the roots of filesystems, System Volume Information and pagefile.sys.
Version 5.1, 1 Aug, 2011:
performance and stability improvements
new: toggling of response to showmount requests.
new: configurable log rotation. Defaults to previous behavior.
new: directory and file caching duration is now configurable.
fix: interoperability fixes with VMware ESXi.
fix: hanging of the server and 100% cpu usage problems are removed.
fix: moving the error log into the install directory.
fix: proper locking around exports and logging operations.
Version 5.0, 25 February, 2010:
Performance improvements
Improved support for Windows Vista/7
New: support for symbolic links
New: support for international filenames (UTF-8)
New: support for mount protocol version 2
New: notification app separate from service that shows log information
New: NLM and NSM debug options
New: added date to timestamp logging
New: option “Set file modification time after every write” allows users to work around inconvenient Window behavior with respect to file modification times. (Normally Windows does not update the mtime on a file until the file is closed. Since Allegro NFS keeps a file open while it is actively being written-to, a Windows program periodically checking the mtime of the file will not be able to tell that the file is actually being modified until the file is closed.) Using this option updates the modification time after every write. As a result there is a performance penalty of 10% and this option is disabled by default.
Improvements to hard link support
Better compatibility between UNIX and Windows-style locking
Improved interoperatibility with several UNIX clients
