Date: Sat, 24 May 1997 10:15:47 +0100 (BST) From: Doug Rabson <dfr@nlsystems.com> To: Michael Smith <msmith@atrad.adelaide.edu.au> Cc: chuckr@Glue.umd.edu, FreeBSD-Hackers@freebsd.org Subject: Re: 3COM stuff Message-ID: <Pine.BSF.3.95q.970524101244.331I-100000@herring.nlsystems.com> In-Reply-To: <199705240810.RAA05616@genesis.atrad.adelaide.edu.au>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 24 May 1997, Michael Smith wrote: > Doug Rabson stands accused of saying: > > > > > > NFS on those cards is _BAD_. If at all possible, do an FTP install. > > > If you must use NFS, make sure the rsize parameter is _small_ - I > > > recommend 1024 as a general rule. > > > > In theory, NFS over TCP should be usable with these cards. > > This is why? I would have thought that an 8k read would result in 8k > of packets, unless the window is somehow reduced through magic > knowledge of the card's receive capabilities...? TCP will break down the 8k result into MTU sized packets and will also be able to re-send in MTU sized packets if one is lost. The problem with the 3c503 is that NFS reads will send a burst of about 6 packets back-to-back which overflows the buffers. When the request times out, it re-sends and gets back another burst which overflows again. In TCP, the re-sending happens at a lower level and should work better with the limited buffering of the card. TCP will treat the buffer overflows as network congestion and back off. -- Doug Rabson Mail: dfr@nlsystems.com Nonlinear Systems Ltd. Phone: +44 181 951 1891 Fax: +44 181 381 1039
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.95q.970524101244.331I-100000>