Date: Sun, 1 Mar 1998 18:58:45 -0500 (EST) From: "John S. Dyson" <dyson@FreeBSD.ORG> To: taob@nbc.netcom.ca (Brian Tao) Cc: dyson@FreeBSD.ORG, freebsd-current@FreeBSD.ORG Subject: Re: Lockup in inode/nfsrcv on loopback NFS mount Message-ID: <199803012358.SAA03679@dyson.iquest.net> In-Reply-To: <Pine.GSO.3.95.980301181316.18657B-100000@tor-adm1> from Brian Tao at "Mar 1, 98 06:19:50 pm"
next in thread | previous in thread | raw e-mail | index | archive | help
Brian Tao said: > On Sun, 1 Mar 1998, John S. Dyson wrote: > > > > There has been a historical problem with NFSv3 hanging on localhost > > filesystems. Is this a possibility? If so, NFSv2 is a short-term > > answer (if it is fast enough for you.) > > Yep, NFSv2 works, but it creeps along very slowly. A dd never > seems to climb above about 400K/sec on a P100, regardless of the block > size. I notice a lot of hard drive chatter during any sustained write > operation. Anyway, I don't actually need loopback NFS mounts... it > was something I tried when tracking down another problem. > > Is it specifically a problem with the lo interface? Can it be > avoided by somehow going through an Ethernet driver? > I think that it is a deadlock mostly at the VFS/NFS level. Using the ethernet driver *might* give enough buffering, but I think that the deadlock is inevitable even in that case. As it is now, NFSv3 on a localhost (or any loopback) will likely be a problem. -- John | Never try to teach a pig to sing, dyson@freebsd.org | it just makes you look stupid, jdyson@nc.com | and it irritates the pig. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199803012358.SAA03679>