Date: Tue, 5 Sep 1995 10:37:47 -0700 (MST) From: Terry Lambert <terry@lambert.org> To: mmead@Glock.COM (matthew c. mead) Cc: hackers@FreeBSD.ORG Subject: Re: any way to limit nfs request throughput? Message-ID: <199509051737.KAA23692@phaeton.artisoft.com> In-Reply-To: <199509041625.MAA06295@Glock.COM> from "matthew c. mead" at Sep 4, 95 12:25:21 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> I have someone that's doing some diskless booting off my machine, > which he really should be purchasing his own drive for. Given that I've > only got a 486/66 with 48M ram and an ne2000 clone ethernet card, his > activity is slowing my machine down a lot. Is there any way to limit the > nfs request frequency, or disallow more than x requests per second to > alleviate some of this problem, or should I just nice -19 the nfs > processes? Any other ideas? Thanks... Nicing down the processes will prevent them from being sheduled to run when another process of higher priority is also ready to run (this is simplistic, and doesn't take into account anti-starvation). As long as the reason you are taking so many requests is actual disk I/O, then nice should work. If he's hitting the heck out of cache, then you can't slow it down because the NFS process isn't sleeping on I/O waits. It's only when a process sleeps that nice takes effect, and only then in deciding how long to delay scheduling it to run in favor of other processes. One way to *significantly* reduce NFS throuput is to load Linux. 8-). Terry Lambert terry@lambert.org --- "It's a joke son, a joke!" -- Foghorn Leghorn --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199509051737.KAA23692>