Date: Sun, 20 Aug 1995 01:23:52 -0700 From: David Greenman <davidg@Root.COM> To: Dmitry Khrustalev <dima@bog.msu.su> Cc: hackers@freefall.freebsd.org Subject: Re: Making a FreeBSD NFS server Message-ID: <199508200823.BAA00409@corbin.Root.COM> In-Reply-To: Your message of "Sun, 20 Aug 95 11:22:59 %2B0400." <Pine.SOL.3.91.950820111605.552D-100000@sunny.bog.msu.su>
next in thread | previous in thread | raw e-mail | index | archive | help
> >On Sat, 19 Aug 1995, David Greenman wrote: > >> >> >> >> > What can I do to improve NFS writes to a FreeBSD server? The >> >> > machine was idle at the time, with four nfsd's running. Should I be >> >> > using UDP or TCP (I turned both on)? >> >> >> > >> >Assuming you are running 2.1 branch: current (2.2) code includes support >> >for write gathering, which can improve large write performance. >> >Be warned: 2.2 kernel is incompatible with 2.1 user space utilities. >> >> If there is interest, I'll implement an NFS_ASYNC kernel option for both >> FreeBSD 2.1 and 2.2. > >Kernel option or per filesystem option? Switching it on for all >filesystems at once is evil. And if nfs3 data corruption problems are >resolved, this option will not have any real value. Initially I was planning on just a kernel option that would enable it for all NFS exports. I could make it per-export by adding an option to /etc/exports, modifying mountd to grok it, passing it to the kernel, etc, but this is starting to become a real project. Regarding NFSv3: v3 is only an option if you have clients that support it. This is actually quite rare and would limit it's use to modern FreeBSD boxes. ...and of course, FreeBSD 2.1 doesn't have v3 and I don't intend to add it before the release. -DG
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199508200823.BAA00409>