Date: Sat, 25 Jan 2014 22:19:23 -0500 (EST) From: Rick Macklem <rmacklem@uoguelph.ca> To: Garrett Wollman <wollman@freebsd.org> Cc: freebsd-fs@freebsd.org, freebsd-net@freebsd.org Subject: Re: Terrible NFS performance under 9.2-RELEASE? Message-ID: <688905116.16333139.1390706363082.JavaMail.root@uoguelph.ca> In-Reply-To: <21220.32074.958702.595502@hergotha.csail.mit.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
Garrett Wollman wrote: > <<On Sat, 25 Jan 2014 21:36:26 -0500 (EST), Rick Macklem > <rmacklem@uoguelph.ca> said: > > > Have you tried increasing readahead by any chance? I think the > > default > > is 1, which means the client will make 2 read requests and then > > wait for > > those replies before doing any more reads. Since you have fast > > links, > > maybe the 2 * 64K reads isn't enough to keep the pipe filled? (This > > depends on latency, which you didn't mention.) > > -o readahead=4 nearly doubles the speed, to a bit over 5 Gbit/s. > And "-o readahead=8" is slower or faster? (I think you can go up to at least 16, but I can't remember the upper bound. It's in one of the .h files.;-) > Oddly, when I unmount the filesystem, the test client sometimes > freezes for 15-30 seconds. Since I'm not on the console I can't tell > what it's doing when this happens. > Hmm, no idea. Maybe it takes a while to throw away all the buffer cache blocks? I run such small systems by to-days standards, I wouldn't see a delay that "might" occur for a large buffer cache. At least a little progress, rick > -GAWollman > > _______________________________________________ > freebsd-net@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-net > To unsubscribe, send any mail to > "freebsd-net-unsubscribe@freebsd.org" >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?688905116.16333139.1390706363082.JavaMail.root>