Footnotes


[1]Network File System (NFS) is believed to be a registered trademark of Sun Microsystems Inc.
[2] Pcnfsd is available in source form from Sun Microsystems and many anonymous ftp sites.
[3]Any security benefit of this is highly questionable and as such the BSD server does not require a client to use a reserved port number.
[4]The Encore Multimax is known to require this.
[5]Due to a network partitioning or similar.
[6]Unfortunately, there are also some resource allocation situations in the BSD kernel where the termination signal will be ignored and the process will not terminate.
[7]At best, an extraneous RPC request retransmit increases the load on the server and at worst can result in damaged files on the server when non-idempotent RPCs are redone [Juszczak89].
[8]6 IP fragments for an Ethernet, which has an maximum transmission unit of 1500bytes.
[9]After the first retransmit timeout, the initial interval is backed off exponentially.
[10]Even 0.1% of the total RPCs is probably significant.
[11]Read/write data sizes greater than 8Kbytes will not normally improve performance unless the kernel constant MAXBSIZE is increased and the file system on the server has a block size greater than 8Kbytes.
[12]There are rumors of commercial NFS over TCP implementations on the horizon and these may well be worth exploring.
BUFPAGES is the number of physical machine pages allocated to the buffer cache. ie. BUFPAGES * NBPG = buffer cache size in bytes
[13]I would appreciate email from anyone who can provide NFS vs. NQNFS performance measurements, particularly fast clients, many clients or over an internetwork connection with a large ``bandwidth * RTT'' product.
[14]Some NFS implementations run with UDP checksums disabled, so garbage RPC messages can be received.
[15]Unfortunately, a monitoring tool that uses these logs is still in the planning (dreaming) stage.
[16] A lease is a ticket permitting an activity that is valid until some expiry time.
[17] Write sharing occurs when at least one client is modifying a file while other client(s) are reading the file.
[18] Cached write data is not yet pushed (written) to the server.
[19] The NFS clients will continue to be approximately consistent with the server.
[20] The last lease expiry time may be safely estimated as "boottime+maximum_lease_term+clock_skew" for machines that cannot store it in nonvolatile RAM.
[21] Gray and Cheriton avoid this problem by using a write through policy.
[22] To be safe, the client may only assume that the lease is valid for ``duration'' seconds from when the RPC request was sent to the server.